I changed URLs to :class:`~scrapy.http.Request` in start_urls explanation of the default spider

Release notes for the upcoming 1.5.0 version 

Add mention to dont_merge_cookies in CookiesMiddlewares docs (#2999)

[MRG+1] Use HTTP pool and proper endpoint key for ProxyAgent

Add PyPy3 support

These built-in functions are exposed as methods in PyPy3.
For scrapy this does not matter as:
1) they do not work for CPython at all
2) get_func_args is checked for presense of an argument in scrapy,
   extra "self" does not matter.
But it still makes sense to leave these tests so that we know we
shouldn't use get_func_args for built-in functions/methods.

[MRG+1] setup.py: mention that we support PyPy. See GH-2213.

[MRG+1] Mention PyPy support, add PyPy to install docs

[MRG+1] Drop support for EOL Python 3.3

[CI skip]

DOC update "Contributing" docs

Thanks @eliasdorneles!

* suggest Stack Overflow for Scrapy usage questions;
* encourage users to submit test-only pull requests with reproducable examples;
* encourage users to pick up stalled pull requests;
* we don't use AUTHORS file as a main acknowledgement source;
* suggest using Sphinx autodocs extension

[MRG+1] Issues a warning when user puts a URL into allowed_domains (#2250)

[MRG+1] Scrapy Command: add --meta/-m to the "parse" command to pass additional meta data into the request

checking for subclass of URLWarning instead of checking error message text when URL in allowed_domains

Fixed spelling mistake from "will be pass" to "will be passed"

[MRG+1] Added from_crawler to middleware docs

Separated import statements

Just separated the import statements. Tiny change - testing GitHub!

[MRG+1] Revise/modernize item exporter example in docs

[MRG+1] Use https link in default user agent