Revert "Add .warmup() for triton.jit()" (#682)

Reverts openai/triton#671

It seems like for some reason this caused out-of-memory errors on some
of our internal workloads. I'm reverting this so that HEAD can be used
in production at OpenAI, and I will work on digging into this issue
asynchronously.

python/test/unit/runtime/test_cache.py

@@ -150,25 +150,3 @@ def test_constexpr_not_callable() -> None:
     except BaseException:
         error = True
     assert error is True
-
-
-def test_jit_warmup_cache() -> None:
-    @triton.jit
-    def kernel_add(a, b, o, N: tl.constexpr):
-        idx = tl.arange(0, N)
-        tl.store(o + idx,
-                 tl.load(a + idx) + tl.load(b + idx))
-
-    args = [
-        torch.randn(32, dtype=torch.float32, device="cuda"),
-        torch.randn(32, dtype=torch.float32, device="cuda"),
-        torch.randn(32, dtype=torch.float32, device="cuda"),
-        32,
-    ]
-    assert len(kernel_add.cache) == 0
-    kernel_add[(1,)].warmup(torch.float32, torch.float32, torch.float32, 32)
-    assert len(kernel_add.cache) == 1
-    kernel_add[(1,)].warmup(*args)
-    assert len(kernel_add.cache) == 1
-    kernel_add[(1,)](*args)
-    assert len(kernel_add.cache) == 1