btf: optimize Spec.Copy

[lmb]

btf: avoid allocating typeDeque in copier.copy

    goos: linux
    goarch: amd64
    pkg: github.com/cilium/ebpf/btf
    cpu: 12th Gen Intel(R) Core(TM) i7-1260P
                │   old.txt   │              new.txt               │
                │   sec/op    │   sec/op     vs base               │
    SpecCopy-16   37.76m ± 2%   33.27m ± 5%  -11.91% (p=0.002 n=6)

                │   old.txt    │               new.txt               │
                │     B/op     │     B/op      vs base               │
    SpecCopy-16   38.10Mi ± 0%   32.28Mi ± 0%  -15.27% (p=0.002 n=6)

                │   old.txt   │              new.txt               │
                │  allocs/op  │  allocs/op   vs base               │
    SpecCopy-16   294.7k ± 0%   203.2k ± 0%  -31.05% (p=0.002 n=6)

Signed-off-by: Lorenz Bauer <lmb@isovalent.com>

btf: pre-size copier

Make sure copier has the appropriate size from the start.

    goos: linux
    goarch: amd64
    pkg: github.com/cilium/ebpf/btf
    cpu: 12th Gen Intel(R) Core(TM) i7-1260P
                │   old.txt   │              new.txt               │
                │   sec/op    │   sec/op     vs base               │
    SpecCopy-16   43.96m ± 8%   37.76m ± 2%  -14.10% (p=0.002 n=6)

                │   old.txt    │               new.txt               │
                │     B/op     │     B/op      vs base               │
    SpecCopy-16   48.44Mi ± 0%   38.10Mi ± 0%  -21.36% (p=0.002 n=6)

                │   old.txt   │              new.txt              │
                │  allocs/op  │  allocs/op   vs base              │
    SpecCopy-16   299.3k ± 0%   294.7k ± 0%  -1.56% (p=0.002 n=6)

Signed-off-by: Lorenz Bauer <lmb@isovalent.com>

btf: add Spec.Copy benchmark

Signed-off-by: Lorenz Bauer <lmb@isovalent.com>

[rgo3]

Newbie question: how did you create those outputs that you put into the commit messages?

[lmb]

That is https://pkg.go.dev/golang.org/x/perf/cmd/benchstat !