feat: enhance memory operations with zeroed allocation and improved e…

…rror handling

chore: refactor disassembler, tests, and utilities

chore: add Decoder type for reusable cgo decode buffers

Introduces a Decoder type that preallocates and reuses cgo-side
C.Instruction buffers across multiple decode operations, reducing
allocation overhead. The existing DecomposeInto and DecomposeBatch
functions now wrap a short-lived Decoder to maintain API
compatibility.

The Engine now embeds a Decoder to avoid per-instruction allocations
during emulation.

feat: add zero-allocation DecomposeInto and batch disassembly APIs

Introduces stack-allocated instruction types (Inst, Op) and batch
processing to eliminate heap allocations in hot paths:

- DecomposeInto(): decode into caller-provided Inst
- DecomposeBatch(): decode multiple instructions in one C call
- Inst/Op: fixed-size types with MAX_OPERANDS/MAX_REGISTERS arrays
- Lazy string formatting via Inst.Disassemble()
- Operation-based dispatch via SupportsOp() interface

Benchmarks show significant allocation reduction for decode-heavy
workloads. Existing Decompose() API unchanged for compatibility.

Update goreleaser.yml

ci: fix goreleaser

Update goreleaser.yml

Update goreleaser.yml

ci: fix goreleaser

Update goreleaser.yml