specs-go: remove x/mod semver dependency

[thaJeztah]

• stacked on specs-go: add test coverage #332

---

specs-go: remove x/mod semver dependency

CDI spec versions are a small, fixed set, so using
golang.org/x/mod/semver just to compare them is unnecessary.

Keep the versions in order instead and use that ordering for validation and
minimum-version checks. This also keeps support for versions with or without a
leading v.

As a side effect, ValidateVersion gets noticeably faster for valid versions
in local benchmarks, with no change in allocations.

I wrote a quick benchmark, but probably not worth to keep (so I can drop that commit);

pkg: tags.cncf.io/container-device-interface/specs-go
                          │ before.txt  │              after.txt              │
                          │   sec/op    │   sec/op     vs base                │
ValidateVersion/first-2     276.9n ± 1%   155.8n ± 1%  -43.71% (p=0.000 n=10)
ValidateVersion/middle-2    279.4n ± 1%   171.1n ± 1%  -38.77% (p=0.000 n=10)
ValidateVersion/last-2      354.9n ± 0%   259.9n ± 1%  -26.78% (p=0.000 n=10)
ValidateVersion/invalid-2   131.9n ± 1%   137.9n ± 1%   +4.55% (p=0.000 n=10)
geomean                     245.3n        175.8n       -28.33%

                          │ before.txt │              after.txt              │
                          │    B/op    │    B/op     vs base                 │
ValidateVersion/first-2     592.0 ± 0%   592.0 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/middle-2    592.0 ± 0%   592.0 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/last-2      672.0 ± 0%   672.0 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/invalid-2   280.0 ± 0%   280.0 ± 0%       ~ (p=1.000 n=10) ¹
geomean                     506.7        506.7       +0.00%
¹ all samples are equal

                          │ before.txt │              after.txt              │
                          │ allocs/op  │ allocs/op   vs base                 │
ValidateVersion/first-2     5.000 ± 0%   5.000 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/middle-2    5.000 ± 0%   5.000 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/last-2      8.000 ± 0%   8.000 ± 0%       ~ (p=1.000 n=10) ¹
ValidateVersion/invalid-2   4.000 ± 0%   4.000 ± 0%       ~ (p=1.000 n=10) ¹
geomean                     5.318        5.318       +0.00%
¹ all samples are equal