without any external wiring.

This patch introduces the new method
Esp.rebootIntoUartDownloadMode()

When the user calls this method the ESP8266 reboots into the UART
download mode. In this mode the user can use esptool.py to flash a new
firmware file. The following command was used to test it:
$ esptool.py --before no_reset --after soft_reset --chip esp8266 \
    --port /dev/ttyUSB0 --baud 460800 write_flash 0x0 firmware.bin

The implementation is based on the original implementation in the
boot ROM. Some parts of the original implementation can be found in
[1]. This patch is a squashed and simplified version of [2]. The non
squashed version might be helpful in case of debugging issues.

[1] https://github.com/twischer/xtensa-subjects/blob/master/reversed/bootrom.c
[2] https://github.com/twischer/Arduino/tree/reboot_uart_download_fullSigned-off-by: NTimo Wischer <twischer@freenet.de>

Improve resume speed by passing in last known BSSID
Provide a simpler example for WIFI_SHUTDOWN/WIFI_RESUME
Add documentation for WIFI_SHUTDOWN and WIFI_RESUME.

Fixes #6805

Update the build.board for several boards which were originally added
with a generic type.

time_t is now 64 bits. String(time_t) was ambiguous
tests added

* Fix ESP8266SdFat architecture

Avoid problems reported in
https://forum.arduino.cc/index.php?topic=726897.msg4889319

* Fix Windows CI, python3 now *maybe* exists

Python3 used to be called "python.exe" on earlier VMs, but it looks like
the image has been updated and a "python3.exe" does now exist.  Update
the CI script to first check it "python3" exists, and if not then do the
copy hack, OTW do nothing.

RODATA can be copied automatically by the bootrom, so no reason not to
allow its use for strings and constants in eboot.c

Revert to pfalcon's original uzlib since the single patch to remove
RODATA is not required.

Rationalize eboot.ld linker script, clean up BSS and init it in code.

Saves 112 bytes of space in the bootloader sector by removing the
extra code associated with literal loads.

* Move CRC out of bootload sector

We added protection to only erase the bootload sector when flashing an
image when the new sector != the old sector.  This was intended to
minimize the chance of bricking (i.e. if there was a powerfail during
flashing of the boot sector the chip would be dead).

Unfortunately, by placing the CRC inside the eboot sector *every*
application will have a unique eboot sector (due to the crc/len), so
this protection doesn't work.

Move the CRC into the first 8 bytes of IROM itself.  This frees up extra
space in the boot sector and ensures that eboot won't be reflashed
unless there really is an eboot change.

Update newlib to enable the __ieee754_remainder(f) calls required by
std::remainder and others.

Add device test for std::remainder variants.

Fixes #7845

Avoid issues like #7847 by downloading and checking SHA for each file in
the package.json bundle.

Fatfingered the 32-bit Windows pointer to the Python interpreter.  Our
CI and my own testing missed due to being on 64-bit Windows.

For 2.7.4 release, I'll add a file of the appropriate name to the
release for now, but this will correct things for 3.0.0 and forward.

A couple board types reported ESP8266_GENERIC instead of their proper
types in boards.txt (and in defined generated therefrom/etc.).

Give them proper board types based on their names, like other modules.

* allow to set pin to OUTPUT_OPEN_DRAIN in analogWrite

* remove parameter with default value

* Update core_esp8266_wiring_pwm.cpp

* update documentation accordingly

Adjust the number of bittimes when returning the real I2S rate.

No functional change.  Make the main core and eboot build using the
same C standard version (gnu17).

Add basic 24 bit mode to the I2S API with a i2s_set_bits() call.
By default 16b mode is still used, but if i2s_set_bits(24) is run
before i2s_begin() then the HW will drive 24-bits of data.  This
data must be left-aligned (i.e. bits 31..8) in 4-byte samples.

Fixes #5244 (the HW doesn't support 8 or 32 bits, only 16 or 24).

* hexdump() must be "C"; add ascii data in dump
* remove previous version

Add support for hidden SSID's Multi WiFi

Make installCertStore virtual so we can inherit from CertStore and override it
Create CertStoreBase to inherit from

* Allow specifying waveform generator in source code

Allows code to explicitly specify which waveform generator it wants,
without needing to use one of the 100 IDE menus or adding a `-D`
compile-time define.

Uses weakrefs to allow for apps to call `enablePhaseLockedWaveform();`
within their `setup()` (or anywhere, really) and have the phase locked
versions override the default waveform generators automatically.

For example:

````
void setup() {
  // Uncomment following line to use phase-locked waveform generator
  // enablePhaseLockedWaveform();
  Serial.begin(115200);
  pinMode(LED_BUILTIN, OUTPUT);     // Initialize the LED_BUILTIN pin as an output
  analogWriteRange(1000);
}
void loop() {
  analogWrite(LED_BUILTIN, 100);
  delay(1000);                      // Wait for a second
  analogWrite(LED_BUILTIN, 900);
  delay(2000);                      // Wait for two seconds (to demonstrate the active low LED)
}
````

Also adds an example showing it's use.

Address @dok-net's comments and also remove the _weak/_bound version of
startWaveform() since it's invariant of the actual waveform generator.

* Replacement for Boot ROM aes_unwrap

* Removed unnecessary test

Remove a straggler `undef min/max` from core header
Supersedes #3856

* Add inline always option to HeapSelect

* Add option to force DRAM for pvPort... APIs

* revert print_loc premature change

* Renamed macro to be more specific, FORCE_ALWAYS_INLINE to FORCE_ALWAYS_INLINE_HEAP_SELECT

Added replacement for the Boot ROM `_xtos_set_exception_handler`
to handle installing our replacement `_xtos_c_wrapper_handler`.

Simplified install in the non 32-bit exception module to make use of the
improved `_xtos_set_exception_handler`

Reorganized and improved comments.

* Avoid UB and abort on nullptr buffer
* Report OOM on return value

Fixes #7814.

Return 0, not MAXINT, when a read is called on a File without a backing
instance of a SPIFFS/LittleFS/SD File.

If a server returns "HTTP/1.x -8 OK", for example, it can misguide an application developer into freeing less-important memory so the request can be retried and succeed, when the problem is in the server.

_returnCode is never used anywhere else, but it could still contain a negative value returned by a broken server and therefore could cause troubles in the future (if _returnCode is in fact used)

Fixes #6410

We actually provide a function prototype for `analogReference()` in
`Arduino.h`, but no implementation.  Add a dummy one that only supports
DEFAULT (like other Arduino boards).

resets. It seems you should not use input registers for scratch registers.
Add an extra output register instead. No code size increase.

Light refactoring for readability

Added "C" reference code for Extended ASM

Save two cycles by loading a0 early in exc-c-wrapper-handler.S

Use optimization O2
Net change in size, 0 bytes with optimization. Save 4 bytes w/o Optimization.
With changes and "O2" save 3 cycles on write and 6 cycles on read.

Make verify fail more visible.

Updated example irammem.info to show performance and test int16 handling
when IRAM is used.

Removed 4K performance test. It didn't show any new information over the 1K test.

Addresses compilation errors with pending PRs due to commit 95fb1045

emulation on host: millis()/micros() now start at 0