Add FuzzTest integration, recreating the existing xmlio test

[seang-ilm]

Integrate Google TestFuzz into MaterialX, and as a proof of concept re-create the existing Xmlio Fuzz Test using resources/Materials/Examples as a starter corpus

./MaterialXFuzz   --replay_corpus=MaterialXXmlIoFuzz.ParseXmlString 
...
Note: Google Test filter = MaterialXXmlIoFuzz.ParseXmlString
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from MaterialXXmlIoFuzz
[ RUN      ] MaterialXXmlIoFuzz.ParseXmlString
FUZZTEST_PRNG_SEED=MW5FQRTxoOIxOHCzmQCr17OSwQlpATxEgm9FetA_TN8
[*] Corpus size:     1 | Edges covered:   1438 | Fuzzing time:       9.201566ms | Total runs:  1.00e+00 | Runs/secs:   108 | Max stack usage:    50688
[*] Corpus size:     2 | Edges covered:   1450 | Fuzzing time:      31.130168ms | Total runs:  2.00e+00 | Runs/secs:    64 | Max stack usage:    50688
...
[*] Corpus size:    47 | Edges covered:   2790 | Fuzzing time:   4m44.48424777s | Total runs:  9.00e+01 | Runs/secs:     0 | Max stack usage:    50688
[*] Corpus size:    48 | Edges covered:   2791 | Fuzzing time:  4m44.531136569s | Total runs:  9.20e+01 | Runs/secs:     0 | Max stack usage:    50688
[       OK ] MaterialXXmlIoFuzz.ParseXmlString (287101 ms)
[----------] 1 test from MaterialXXmlIoFuzz (287101 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (287101 ms total)
[  PASSED  ] 1 test.

Note we've got some inputs in the resources/Materials/Examples that are very big which is why we see slow "Runs/secs" in the fuzzer. And you'll see some warnings about some test inputs in that folder that are too big for limit we've set in the Fuzzer of 1M inputs (/MaterialXFuzz/XmlIo_fuzz.cpp:22: INVALID_ARGUMENT: Invalid value in aggregate >> Invalid size: 1671504. Max size: 1048576:)

Sharing for initial feedback.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: seang-ilm / name: Sean Gilligan (1d2f65f)

[seang-ilm]

Pushing a fix to only seed the fuzzer with .mtlx documents. Including other large files in the examples/ dir like the images causes some very slow running. While fuzzing from valid XML I'm seeing steady 1,300 run/s second for a core while ASan in enabled.

[*] Corpus size:  1752 | Edges covered:   3916 | Fuzzing time:  9m56.837074352s | Total runs:  7.81e+05 | Runs/secs:  1308 | Max stack usage:     8368

[.] Fuzzing was terminated.

=================================================================
=== Fuzzing stats

Elapsed time: 10m0.001227882s
Total runs: 788050
Edges covered: 3916
Total edges: 96670
Corpus size: 1752
Max stack used: 8368

This approach still has some drawbacks. This is currently using random mutation on the initial document set, so you burn a lot of cycles trying to generate valid XML. In future this would benefit from a more detailed FuzzTest - Domain that is more likely to generate valid XML more often. Currently the fuzzer has been mainly useful for finding slow input paths than actual security issues.