btcctl: preserve semantic version separators

[pucedoteth]

65db493 ("version: preserve semantic version separators", #2578) added periods to semanticAlphabet in the root version.go, because SemVer separates pre-release identifiers with periods and the old filter silently collapsed beta.rc1 into betarc1.

cmd/btcctl carries its own near-identical copy of that file, and it was not updated. The two now disagree:

version.go:14            ...uvwxyz-."
cmd/btcctl/version.go:14 ...uvwxyz-"

So btcctl still drops the separator. The upstream test, run unchanged against the btcctl copy, fails:

--- FAIL: TestNormalizeVerString
    version_test.go:16: unexpected normalized version: got "betarc1", want "beta.rc1"

This is reachable rather than latent. normalizeVerString is applied to appBuild, which the comment above it documents as settable at build time:

// appBuild is defined as a variable so it can be overridden during the build
// process with '-ldflags "-X main.appBuild foo' if needed.

SemVer build metadata is dot-separated (exp.sha.5114f85 is an example from the spec itself), so a btcctl built with -ldflags "-X main.appBuild=exp.sha.5114f85" reports 0.23.1-beta+expsha5114f85 — the dots stripped and the metadata no longer round-trips. btcd built the same way reports it correctly.

The change is the one-character fix from 65db493 applied to the btcctl copy, plus that commit's test mirrored into cmd/btcctl, with a build metadata case added for the appBuild path.

go build ./..., go test ./cmd/..., gofmt and go vet are all clean.

I left the stale appMinor/appPatch in the btcctl copy alone (0.23.1 there vs 0.26.2 at the root) since bumping those is a release decision, not part of this fix.

---

Disclosure: written with AI assistance; the analysis, fix and test were verified locally.

[pucedoteth]

Pushed a test-only follow-up.

The original table asserted normalizeVerString(x) == x for every case, which an identity function satisfies — so it could not have failed if the filter were dropped. (Thanks to @Lrifton92 for spotting this on the equivalent btcwallet PR.) The table is now in/want-based with a case whose input contains a character outside the semantic alphabet.

Verified by neutralising the alphabet check in place (if true || strings.ContainsRune(...), which still compiles): the new case fails with got "beta!rc1", want "betarc1". The dotted cases still fail in the other direction if . is removed from the alphabet, so the test now pins both ends.