PolyObject crush hotfix

See merge request STJr/SRB2!506

Use P_MobjInsidePolyobj instead of P_MobjTouchingPolyobj, so that you can be crushed by the polyobject's middle rather than just its edges

PolyObject fixes backport

See merge request STJr/SRB2!505

After looking at the FOF part of P_LineOpening for a while I now realise many of these variables aren't even necessary, so I removed them all.

(Naturally I did the same to the camera equivalent)

# Conflicts:
#	src/p_maputl.c

Edit a lot of the rest of the polyobject-related code in P_LineOpening to make more sense and be more optimised.

* If you collide with a line belonging to a polyobject, you should NEVER have to care about any FOFs that might be present in either sector of the linedef. This could lead to colliding with ghostly FOFs that aren't actually there or something dumb, if someone decided to give either of the polyobject's control sectors FOFs for some reason. We don't want that, obviously.
* Polyobjects without POF_CLIPPLANE apparently are supposed to have a top and bottom "physical" height of value INT32_MAX and _MIN respectively, according to P_CheckPosition ...let's be consistent with this.
* Finally, there is no more need for that back = front nonsense hack anymore with my changes made.

# Conflicts:
#	src/p_maputl.c

P_LineOpening: set int32 max/min as defaults for opentop, openbottom etc if a linedef you touched belongs to a polyobjetc. the only thing that really matters in this scenario is the polyobject itself after all!

# Conflicts:
#	src/p_maputl.c

Modify P_CheckSector with a modified version of Sal's attempted proper fix for polyobjects crushing, so that we only need to check the polyobject's control sector directly in the waypoints code.

This time I've definitely fixed that teleport to ground issue I'm pretty sure, I don't get it in my tests at least.

This reverts commit 749c2753.

Fix crash with GME sounds when being freed

See merge request STJr/SRB2!504

# Conflicts:
#	src/r_data.c

unfuck awayview

See merge request STJr/SRB2!493

Fade out music at end of level, before intermission

See merge request STJr/SRB2!472

Pk3 folders are not textures!

See merge request STJr/SRB2!502

More Memory Shenanigans

See merge request STJr/SRB2!501

# Conflicts:
#	src/r_data.c

Keep the warning though.

A lower size could technically be valid, but could easily run into strange issues.

As far as I know it's basically unused, and the strstr is inherently unsafe because there's no guarantee that a patch's contents are NULL terminated.

Could this be changed to only read the first so many bytes?

Prevent just anyone from skipping credits but allow admins

See merge request STJr/SRB2!478

Pk3 Leaks

See merge request STJr/SRB2!500