Refactor BspLump

[snake-biscuits]

Editing attrs of an object returned by a BspLump don’t update BspLump.changes, need to do some soft copies probably
Slice overriding & insertion don’t seem to work either.
Both of these need tests and more

>>> BspLump[i].x = new_val
>>> BspLump[i].x
old_val

editing a lump should be as easy as modifying the contents of a list

[snake-biscuits]

Need to let copied list elements behave as if BspLump classes are real lists
Changes can be made at any time to any touched return value
Slices should return copies

__setitem__(self, index, value) should allow insertion when index is a slice
Tracking insertions with minimal changes stored would be ideal (track insertions and offset indices for __getitem__)

each BspLump type will need a test for this, do not have RespawnBsp bsp_lumps to test
Titanfall2 repo has r2 mp_lobby so use that

[snake-biscuits]

Changes can be made at any time to any touched return value

This wouldn't work with bytes objects, as they are immutable, but allowing changes to RawBspLump would be nice

[snake-biscuits]

should really address this before tackling #15 as making small edits & seeing their affects in-engine is very powerful for research
but when the bsp_tool implementation differs from user expectations you're gonna see a lot of bugs & wacky workarounds

gotta love writing my own kernel to handle big data

[snake-biscuits]

gotta love writing my own kernel to handle big data

No, really, we need to do garbage collection of some kind here
Each reference to an object collected will need to feed back into the changes attribute of the BspLump it came from

bsp.VERTICES[0].z += 1 # doesn't work rn. but should
{print(v) for v in bsp.VERTICES[:64]}  # makes a copy and deletes it within scope; no changes possible

Maybe we could override the __del__ method of each LumpClass instance to remove itself from changes if unchanged?

We would also need to accurately mimic how Python creates copies & deep-copies
Ideally we hook existing Python memory management systems to handle all of this for us.
Some people smarter than us may have already tackled this, idk.
Could be a bunch of talks on messing with memory management & working with large files, I haven't found them yet tho.

The odds someone smarter than me has already tried this are quite high.
Research time should be put into finding examples tackling this problem.