Add Chunk class & template_heap_command

[CptGibbon]

The Chunk class abstracts developers from some heap & gdb internals and makes use of the gdb module's lazy fetching.
The template_heap_command command serves as an example of how to use the Chunk class (and hopefully future classes).

[codecov-commenter]

Codecov Report

Merging #1159 (dfdcd20) into dev (96d90d1) will increase coverage by 1.13%.
The diff coverage is 27.02%.

@@            Coverage Diff             @@
##              dev    #1159      +/-   ##
==========================================
+ Coverage   51.45%   52.59%   +1.13%     
==========================================
  Files         176      178       +2     
  Lines       19997    20118     +121     
  Branches     1833     1851      +18     
==========================================
+ Hits        10290    10581     +291     
+ Misses       9325     9129     -196     
- Partials      382      408      +26     

Impacted Files	Coverage Δ
pwndbg/heap/ptmalloc.py	37.51% <22.22%> (-0.94%)	⬇️
pwndbg/commands/heap.py	60.42% <40.00%> (-0.57%)	⬇️
tests/test_commands_next.py	95.65% <0.00%> (ø)
tests/test_mprotect.py	100.00% <0.00%> (ø)
pwndbg/commands/__init__.py	69.68% <0.00%> (+0.45%)	⬆️
pwndbg/proc.py	75.00% <0.00%> (+0.58%)	⬆️
pwndbg/commands/telescope.py	80.41% <0.00%> (+1.03%)	⬆️
pwndbg/commands/nearpc.py	88.52% <0.00%> (+1.63%)	⬆️
pwndbg/lib/abi.py	95.65% <0.00%> (+2.17%)	⬆️
pwndbg/commands/vmmap.py	43.47% <0.00%> (+2.17%)	⬆️
... and 21 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[lebr0nli]

I think this PR is great, the only tiny potential issue I can find is: should we use f-strings syntax?
I like f-strings and it's convenient, but for the compatibility, maybe we should not use it until we fully deprecate Python<3.6 support I guess?
(I saw the comment from @disconnect3d about the Python version we want to support here: #1063 (comment))

[gsingh93]

maybe we should not use it until we fully deprecate Python<3.6 support I guess?

We had a discussion on Discord after that comment, I remember the general consensus was we should support distros that are not EOL yet, which would include Ubuntu 18.04, 20.04 and 22.04, and we now run the CI on all three. Ubuntu 18.04 uses Python 3.6.9, so personally I think it's OK to drop support for versions older than 3.6, but I know we always have the issue of someone wanting to use it on those older versions.

[CptGibbon]

So far the consensus seems to be:

1. Ditch template_heap_command
2. Keep f-strings

@gsingh93 are you happy to postpone the unit testing stuff?
@disconnect3d do you agree with the above?

[gsingh93]

Yup, feel free to hold off on the unit testing stuff, we can revisit it later.

[disconnect3d]

Yeah, I am fine with that. So what with the Chunk class? Are you going to refactor heap commands to use it?:P

[CptGibbon]

So what with the Chunk class? Are you going to refactor heap commands to use it?

Exactly. Right now contributors must manually find many heap values in their commands:

pwndbg/pwndbg/commands/heap.py

Lines 672 to 674 in 1ca4d2d

	size_field = pwndbg.gdblib.memory.u(cursor + ptr_size)
	real_size = size_field & ~allocator.malloc_align_mask
	prev_inuse = allocator.chunk_flags(size_field)[0]

Hopefully we can avoid this (and benefit from the built-in efficiency the gdb module gives us) by introducing these classes to pwndbg's heap inspection engine.