[CompilerFlag] Detect if FlashInfer is enabled from libinfo (#1941)

This PR supports the detection of if FlashInfer is enabled when
building TVM, so that FlashInfer won't be enabled when TVM is
not built with FlashInfer enabled.

python/mlc_llm/interface/compiler_flags.py

@@ -1,8 +1,11 @@
 """Flags for overriding model config."""
+
 import dataclasses
 from io import StringIO
 from typing import Optional
 
+import tvm
+
 from mlc_llm.support import argparse, logging
 from mlc_llm.support.config import ConfigOverrideBase
 
@@ -65,6 +68,8 @@ def _flashinfer(target) -> bool:
                 return False
             if target.kind.name != "cuda":
                 return False
+            if tvm.get_global_func("support.GetLibInfo")()["USE_FLASHINFER"] != "ON":
+                return False
             arch_list = detect_cuda_arch_list(target)
             for arch in arch_list:
                 if arch < 80: