update llvm to support latest 1250 features.

Release: bump version to 0.3.1 (#988)

1.robust layout done; 2. gfx1250 features; 3. api refine

[compat] Fix global_prefetch scope param: positional → keyword

global_prefetch's second param changed from positional `scope` to
keyword-only `cache_policy` in the LLVM binding regeneration.
l2_prefetch_tile was the only call site missed — fix it to use
cache_policy=_cache_policy_attr(scope), consistent with the
tensor_load_to_lds/tensor_store_from_lds fixes in the same file.

Co-Authored-By: Claude <noreply@anthropic.com>

1. gfx1250:tdm atoms and mma atoms. 2. update doc

1.type promote align; 2.autotune fix; 3.kernel refactor

fix closure types; fix cpointer perf

fix profiler loc; add notebook; opt launch; opt attn

v0.2.0 ready: fix cache key issues; ast_while ready; fix att trace co…

…de map; enable gfx1151

1, refactor cluster.py 2, refactor auto dynamic in if_ 3, change back…

…end config