[IME] Add sliding-window matrix multiply-accumulate instructions

[wujing215]

Summary

This PR adds support for fixed and dynamic sliding-window matrix multiply-accumulate instructions to the IME dialect.

Changes

New Instructions

Fixed Sliding-Window Integer Instructions (12 total)

Slide	Signed×Signed	Unsigned×Unsigned	Signed×Unsigned	Unsigned×Signed
1	vmadot1	vmadot1u	vmadot1su	vmadot1us
2	vmadot2	vmadot2u	vmadot2su	vmadot2us
3	vmadot3	vmadot3u	vmadot3su	vmadot3us

Fixed Sliding-Window Floating-Point Instructions (3 total)

• vfmadot1, vfmadot2, vfmadot3

Dynamic Sliding-Window Instructions (5 total)

• vmadotn, vmadotnu, vmadotnsu, vmadotnus, vfmadotn

Testing

# Build and test assembly generation
cd examples/IMEDialect
make vmadot1-asm    # Generates: vmadot1 v20, v8, v16
make vmadot2u-asm   # Generates: vmadot2u v20, v8, v16
make vmadot3su-asm  # Generates: vmadot3su v20, v8, v16