AdvSimd shift+widen invalid imm8 range

[ovska]

Description

AdvSimd.ShiftLeftLogicalWideningLower and friends have the imm argument limited to one too low. For example only 7 when widening bytes. Docs say you should be able to use the full bit width (e.g. 8 on uint8):
https://developer.arm.com/documentation/dui0473/m/neon-instructions/vshll--by-immediate-

Reproduction Steps

AdvSimd.ShiftLeftLogicalWideningLower(Vector64<byte>.One, 8)

Encountered when trying to widen a byte vector's MSB's to uints.

Expected behavior

Imm8 range 0..8

Actual behavior

Imm8 range 0..7, using 8 produces a diagnostic and throws an exception on runtime

Regression?

Known Workarounds

Widen first, then shift manually (a few extra ops).

Configuration

.NET 10.0.100
macOS 26.1 (25B78)

Other information

No response

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-runtime-intrinsics
See info in area-owners.md if you want to be subscribed.