Initial extraction from isis: direct /dev/kfd GPU compute

AMD RDNA3 GPU driver via raw KFD ioctls. No ROCm, no OpenCL.

- Device discovery and topology queries
- VRAM/GTT/Userptr memory with shared address space
- PM4 compute queue and kernel dispatch
- Hand-written RDNA3 ASM kernels: matmul (3100 GFLOP/s), matvec, superlinear
- Tile-safe buffer padding for OOB protection
- Event-based interrupt wait (no CPU polling)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

target/debug/.fingerprint/kfd-c036a4411ecb62c5/lib-kfd.json

@@ -0,0 +1 @@
+{"rustc":18276270781310494267,"features":"[]","declared_features":"[]","target":10069803095152535176,"profile":17672942494452627365,"path":10763286916239946207,"deps":[[12111499963430175700,"libc",false,1692664963012730200]],"local":[{"CheckDepInfo":{"dep_info":"debug/.fingerprint/kfd-c036a4411ecb62c5/dep-lib-kfd","checksum":false}}],"rustflags":[],"config":8247474407144887393,"compile_kind":0}