Sentinel: Improve INFO command behavior

Improve consistency of INFO MEMORY

Fix quicklist Pop()

Fixes issue #2392.

It's hard to pick a good approach here. A few arguments:

1) There are many exposed instances on the internet.
2) Changing the default when "bind" is not given is very dangerous,
   after an upgrade the server changes a fundamental behavior.
3) Usually Redis, when used in a proper way, will be protected *and*
   accessed often from other computers, so this new default is likely
   not what most people want.
4) However if users end with this default, they are using the example
   redis.conf: likely they are reading what is inside, and they'll see
   the warning.

Now the tests actually compare return values instead of just
verifying _something_ got returned.

Closes #2398

Readme fix

Apparently no refernece-style links supported in Github markdown.

Less grays: more readable palette since usually we have a non linear
distribution of percentages and very near gray tones are hard to take
apart. Final part of the palette is gradient from yellow to red. The red
part is hardly reached because of usual distribution of latencies, but
shows up mainly when latencies are very high because of the logarithmic
scale, this is coherent to what people expect: red = bad.

Unify to uppercase the headline

Related to PR #2357.

Support "1G" etc. units in CONFIG SET maxmemory

Related to PR #2357.

The old version of SPOP with "count" argument used an API call of dict.c
which was actually designed for a different goal, and was not capable of
good distribution. We follow a different three-cases approach optimized
for different ratiion between sets and requested number of elements.

The implementation is simpler and allowed the removal of a large amount
of code.

Now the API automatically creates its argv copy and increment ref count
of passed objects.

Severan problems are addressed but still a few missing.
Since replication of this command was more complex than others since it
needs to replicate multiple SREM commands, an old API able to do this
was reused (it was taken inside the implementation since it was pretty
obvious soon or later that would be useful). The API was improved a bit
so that now a command may opt-out for the standard command replication
when the server.dirty counter is incremented, in order to "manually"
replicate what it wants.

Fixed by @oranagra, thank you.

Avoid code repetition introduced with PR #2367, also fixes the return
value to always return 0 if there is nothing more to rehash.