vector2 support for Surface.get_at and set_at

[ScriptLineStudios]

the ability to use vector2's to specify coordinates for surface set_ats and get_ats

[oddbookworm]

I think you should edit the typestub info in buildconfig/stubs/pgyame/surface.pyi so that linters don't throw a fit when using a Vector2 in these methods. Also, adding documentation for this new functionality would be a good idea IMO. Maybe make sure to mention that any floating point values will be truncated, not rounded

[novialriptide]

@oddbookworm has mentioned that linters would complain about introducing Vector2s to get_at() and set_at(), this got me thinking that I'm not entirely sure if this PR should go through as stuff like this becomes valid.

# this pr
import pygame
a = pygame.Surface((50,50))
a.set_at(pygame.Vector2(50,50.5), (25,25,25))
a.set_at((50,50.5), (25,25,25))

whereas before, this would've thrown an error

# pygame-ce 2.2.1
import pygame
a = pygame.Surface((50,50))
a.set_at((50,50.5), (25,25,25))

TypeError: 'float' object cannot be interpreted as an integer

If we favor introducing this, we can mention in the docs that we now support float values for get_at() and set_at().

[narilee2006]

I agree

[Starbuck5]

this got me thinking that I'm not entirely sure if this PR should go through as stuff like this becomes valid

Many things in pygame support floats now, and didn't always. Transform.scale used to need integer dimensions, for example. I think it's perfectly fine to have integer things accept floats and just truncate them.

But you're right, this can be documented and hinted as support for a 2-sequence of floats.

[Starbuck5]

Also I'm curious what kind of performance impact this has on get_at/set_at.

[Starbuck5]

I did some performance testing, and set_at with this PR is 1.5% faster now! (This is pretty low, it could be a fluke)
But this means that there isn't a big performance regression here we're not noticing.

I'm not sure whether pg_TwoIntFromObj or that PyArg_ParseTuple is more expensive, but METH_O is less expensive than METH_VARARGS so maybe that helps it.

[oddbookworm]

LGTM! (After a quick merge conflict fix)

[ankith26]

The docs, tests and stubs for get_at_mapped aren't updated yet

[Starbuck5]

Looks good!

Thanks for working on this. Getting stuff supporting floats when appropriate is a good initiative throughout the pygame-ce API.

Changes were made