[llvm] [LV][EVL] Support cast instruction with EVL-vectorization (PR #108351)

[LLVM Continuous Integration via llvm-commits]

llvm-ci wrote:

LLVM Buildbot has detected a new failure on builder `llvm-clang-x86_64-expensive-checks-debian` running on `gribozavr4` while building `llvm` at step 6 "test-build-unified-tree-check-all".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/16/builds/10466

<details>
<summary>Here is the relevant piece of the build log for the reference</summary>

```
Step 6 (test-build-unified-tree-check-all) failure: test (failure)
******************** TEST 'LLVM :: Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll' FAILED ********************
Exit Code: 1

Command Output (stderr):
--
RUN: at line 2: /b/1/llvm-clang-x86_64-expensive-checks-debian/build/bin/opt -passes=loop-vectorize -debug-only=loop-vectorize  -force-tail-folding-style=data-with-evl  -prefer-predicate-over-epilogue=predicate-dont-vectorize  -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-max=128 -disable-output < /b/1/llvm-clang-x86_64-expensive-checks-debian/llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll 2>&1 | /b/1/llvm-clang-x86_64-expensive-checks-debian/build/bin/FileCheck --check-prefix=IF-EVL /b/1/llvm-clang-x86_64-expensive-checks-debian/llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll
+ /b/1/llvm-clang-x86_64-expensive-checks-debian/build/bin/opt -passes=loop-vectorize -debug-only=loop-vectorize -force-tail-folding-style=data-with-evl -prefer-predicate-over-epilogue=predicate-dont-vectorize -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-max=128 -disable-output
+ /b/1/llvm-clang-x86_64-expensive-checks-debian/build/bin/FileCheck --check-prefix=IF-EVL /b/1/llvm-clang-x86_64-expensive-checks-debian/llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll
/b/1/llvm-clang-x86_64-expensive-checks-debian/llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll:9:16: error: IF-EVL-NEXT: is not on the line after the previous match
; IF-EVL-NEXT: Live-in vp<[[VFUF:%[0-9]+]]> = VF * UF
               ^
<stdin>:431:1: note: 'next' match was here
Live-in vp<%0> = VF * UF
^
<stdin>:266:60: note: previous match ended here
VPlan 'Final VPlan for VF={vscale x 1,vscale x 2},UF={1}' {
                                                           ^
<stdin>:267:1: note: non-matching line after previous match is here
Live-in ir<%11> = VF * UF
^

Input file: <stdin>
Check file: /b/1/llvm-clang-x86_64-expensive-checks-debian/llvm-project/llvm/test/Transforms/LoopVectorize/RISCV/vplan-vp-cast-intrinsics.ll

-dump-input=help explains the following input dump.

Input was:
<<<<<<
        .
        .
        .
      426: LV: Scalarizing: %gep = getelementptr inbounds i32, ptr %b, i64 %iv 
      427: LV: Scalarizing: %gep4 = getelementptr inbounds i64, ptr %a, i64 %iv 
      428: LV: Scalarizing: %iv.next = add nuw nsw i64 %iv, 1 
      429: LV: Scalarizing: %exitcond.not = icmp eq i64 %iv.next, %N 
      430: VPlan 'Initial VPlan for VF={1},UF={1}' { 
      431: Live-in vp<%0> = VF * UF 
next:9     !~~~~~~~~~~~~~~~~~~~~~~~  error: match on wrong line
      432: Live-in vp<%1> = vector-trip-count 
      433: Live-in vp<%2> = backedge-taken count 
      434: Live-in ir<%N> = original trip-count 
      435:  
      436: vector.ph: 
        .
        .
        .
>>>>>>

--

...

```

</details>

https://github.com/llvm/llvm-project/pull/108351