feat(lang): string_split/replace/slice, int/float_parse, bool_to_string, 7 map built-ins

[escapeboy]

Summary

• Adds 13 new built-in operations continuing the __builtin_* pattern (opcodes 0x9A–0xA6)
• String: string_split, string_replace, string_slice, int_parse, float_parse, bool_to_string
• Map: map_get, map_set, map_remove, map_contains_key, map_keys, map_values, map_len
• Value::Map is BTreeMap<String, Value> — all map keys are String, determinism guaranteed

Test plan

• 21 new tests added in crates/llmvm/src/tests.rs
• cargo test -p boruna-vm — 187 tests pass
• cargo test -p boruna-compiler — all pass
• cargo clippy --workspace -- -D warnings — clean

🤖 Generated with Claude Code

[github-actions]

Bench compare

Threshold for regression: ≥ 10% slower mean.

Benchmark	Mean change	99% CI
compile_crud_admin_template	-2.99%	[-3.31%, -2.75%]
compile_medium_program	-2.41%	[-2.75%, -2.19%]
compile_small_program	-5.54%	[-5.87%, -5.20%]
evidence_build_5_steps	+0.32%	[-2.33%, +2.31%]
evidence_build_empty	-0.16%	[-0.85%, +0.66%]
evidence_verify_10_steps	+1.08%	[-0.18%, +2.45%]
evidence_verify_5_steps	+2.16%	[-0.84%, +4.46%]
vm_call_dispatch_loop/iters=1000	+3.34%	[+2.94%, +3.59%]
vm_call_dispatch_loop/iters=10000	+4.03%	[+3.96%, +4.10%]
vm_pure_loop/iters=1000	-0.20%	[-0.80%, +0.20%]
vm_pure_loop/iters=10000	-0.39%	[-0.47%, -0.31%]
vm_pure_loop/iters=100000	+0.35%	[+0.14%, +0.63%]
vm_record_loop/iters=1000	+4.77%	[+4.65%, +4.88%]
vm_record_loop/iters=10000	+4.32%	[+4.09%, +4.53%]

✅ No benchmark regressed past threshold.