mdlpp: add animation, sequence, flex, and IK support

[cueki]

put head in wood chipper for 1 week see what comes out

New

Added support for bone animation data with compressed quaternions/vectors and RLE decompression, animation descriptors with frame count/fps/flags, IK rules/chains/links and error data, and movement/velocity data.

Sequence parsing now includes descriptors with blending and timing, animation events, auto layers, IK locks, activity names/weights, keyvalue strings, and pose parameters.

Flex/facial system includes eyeball movement parameters, vertex animations, flex controllers/UI config/rules/operators (all flex operation types), mesh flex controllers, and mouth bone references.

Also added all 21 bone flags, attachment points on bones, external model and animation block references, and basic Header2 parsing.

Tested with the included v44, v47,and v49 MDL files and my own v48s (TF2 Heavy and Bot Heavy models).

WIP

• External animation files not yet supported (only embedded animations with animBlock == 0)
• Some animation features not parsed: local hierarchy, animation sections, zero frame data
• Some sequence offset fields read but arrays not parsed: animIndexIndex, weightListIndex, poseKeyIndex, cyclePoseIndex
• Header2 arrays for source bone transforms, linear bone table, and bone flex drivers not fully parsed
• Untested: local node transitions, model-local attachments, global IK autoplay locks (didn't have in models)

[cueki]

Tested against all models from HL2, TF2, CS:GO, Portal 2, and L4D2. 30,456 succeeded, 38 failed. The failures are 2 corrupted TF2 files, 2 CS:GO hash mismatches, and 34 L4D2 models that seem to use a different vtx strip group format.

[craftablescience]

oh wow L4D2 parses? I didn't expect it to have the same format

[cueki]

with 10x as many errors comparably :), I thought I'd give it a shot anyways... it would require some more work to actually be fully 'supporting' the format

[SpringierTrain]

does it work with HL2X

[craftablescience]

Don't worry too much about fixing these, I will get around to it eventually

[cueki]

Should be good now :)

Not entirely confident about the Math.h change, I think it's what you wanted.

[cueki]

we can probably remove this comment here and at #R459 ...

unless its something specific you care about, the whole header is a parsed representation