[Based on #103] debug: add LLDB goroutine backtraces

[cpunion]

Based on #103.

Adds goroutine-aware native backtraces to the LLGo LLDB adapter:

• records each LLGo M OS thread ID using Darwin pthread_threadid_np or Linux SYS_GETTID, matching Go runtime m.procid semantics
• maps live G/M/P records to LLDB threads without changing the selected thread
• adds llgo goroutine <id> bt / backtrace, bounded to 256 frames
• reports unavailable mappings cleanly for unsupported runtime layouts and targets

Validation:

• macOS arm64, Apple LLDB 1703: 276/276 integration assertions
• Linux arm64, LLDB 19.1.7: 276/276 integration assertions
• runtime G isolation, G/M/P links, allg lifecycle, and OS thread-ID execution tests on both platforms
• standard DWARF checks at O0, O1, O2, O3, Os, and Oz
• cmd/internal/lldb statement coverage: 87.5%

Part of xgo-dev#2164 (native debugger Task 3).

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

[github-actions]

LLGo baseline benchmarks

df1a0185c53b | workflow run | long-term charts

Program measurements

Platform	Workload	File size	vs main	Build	vs main	Run	vs main
Linux	cprintf	19344 B	+4.3% (worse)	362.373 ms	-12.1% (better)	1.298 ms	-16.3% (better)
Linux	fmtprintf	2218824 B	+0.2% (worse)	3.379 s	-12.4% (better)	2.451 ms	-25.4% (better)
Linux	println	73832 B	+1.6% (worse)	376.435 ms	-5.9% (better)	1.583 ms	-5.4% (better)
macOS	cprintf	84672 B	+0.0%	488.231 ms	+17.8% (worse)	4.049 ms	+12.8% (worse)
macOS	fmtprintf	2361696 B	+0.7% (worse)	4.017 s	+2.0% (worse)	25.844 ms	+10.6% (worse)
macOS	println	125888 B	-0.9% (better)	478.700 ms	+15.5% (worse)	5.250 ms	-27.7% (better)

Core language and compiler benchmarks

Platform	Benchmark	ns/op	vs main
Linux	BenchmarkLookupPCRandom	13.250 ns/op	-1.7% (better)
Linux	BenchmarkMergeCompilerFlags	151.200 ns/op	-2.1% (better)
Linux	BenchmarkMergeLinkerFlags	95.220 ns/op	-6.0% (better)
Linux	BenchmarkChannelBuffered	34.910 ns/op	+0.4% (worse)
Linux	BenchmarkChannelHandoff	28252 ns/op	-11.8% (better)
Linux	BenchmarkDefer	44.210 ns/op	-21.5% (better)
Linux	BenchmarkDirectCall	1.564 ns/op	+0.1% (worse)
Linux	BenchmarkGlobalRead	1.557 ns/op	-0.2% (better)
Linux	BenchmarkGlobalWrite	2.480 ns/op	-0.6% (better)
Linux	BenchmarkGoroutine	43372 ns/op	-42.6% (better)
Linux	BenchmarkInterfaceCall	8.102 ns/op	+4.0% (worse)
Linux	BenchmarkRuntimeGetG	2.182 ns/op	-58.9% (better)
macOS	BenchmarkLookupPCRandom	14.370 ns/op	+15.5% (worse)
macOS	BenchmarkMergeCompilerFlags	157.500 ns/op	+26.5% (worse)
macOS	BenchmarkMergeLinkerFlags	119.500 ns/op	+72.3% (worse)
macOS	BenchmarkChannelBuffered	29.240 ns/op	+5.4% (worse)
macOS	BenchmarkChannelHandoff	13108 ns/op	+87.0% (worse)
macOS	BenchmarkDefer	46.220 ns/op	+23.5% (worse)
macOS	BenchmarkDirectCall	1.255 ns/op	-13.6% (better)
macOS	BenchmarkGlobalRead	1.317 ns/op	+8.8% (worse)
macOS	BenchmarkGlobalWrite	1.461 ns/op	-1.7% (better)
macOS	BenchmarkGoroutine	49602 ns/op	+56.7% (worse)
macOS	BenchmarkInterfaceCall	6.570 ns/op	+21.7% (worse)
macOS	BenchmarkRuntimeGetG	2.783 ns/op	+8.0% (worse)

Compared only with the latest matching platform in the main series.

Warning

• Persistent publishing is unavailable because no data token is configured.