alpha: sign-extend memory-format displacement

[Samin061]

Your checklist for this pull request

• I've documented or updated the documentation of every API function and struct this PR changes.
• I've added tests that prove my fix is effective or that my feature works (if possible)

Detailed description

The Alpha memory-format instructions (lda, the integer loads/stores, the store-conditional forms, and the lds/ldt/sts/stt float loads and stores) carry a 16-bit displacement that the ISA treats as signed, but the decoder read it out of the instruction word as a plain unsigned field and never sign-extended it, so any displacement with bit 15 set decoded to a large positive value. cstool -d alpha 0xb0,0xb0,0x42,0x20 shows it: the offset prints as 0xb0b0 and operands[1].imm is 0xb0b0, where it should be -20304. The displacement is read in the shared decode macro at the four memory-format cases (indices 1, 22, 23, 24), each storing fieldFromInstruction_4(insn, 0, 16) straight into the operand. Casting that field through int16_t at those four sites stores the signed offset, so the same word now decodes to lda $2,-0x4f50($2). Positive displacements are unaffected.

Test plan

cstool -d alpha 0xb0,0xb0,0x42,0x20 now prints lda $2,-0x4f50($2) with operands[1].imm = 0xffffffffffffb0b0. I added negative-displacement cases to tests/MC/Alpha/insn-alpha.s.yaml covering the integer, store-conditional and float-memory forms, and updated the two detail cases in tests/details/alpha.yaml and tests/details/cs_common_details.yaml that had captured the old unsigned value.

Closing issues

closes #3021