feat: add slim binary release variant

[apstndb]

Summary

• add cmd/spanner-mycli-slim, a second main package that registers no optional features
• keep the root binary as the full variant with GEMINI/LLM, BigQuery, and CQL
• publish distinct full and slim GoReleaser builds and archives
• test full/slim statement registration and document source/release builds
• document a reproducible controlled relative-size comparison

This uses separate main packages rather than build tags, so ordinary go test ./..., vet, and lint continue to compile both variants. It is stacked on the GoReleaser scaffold cleanup in #789.

The primary rationale is least capability: users who do not need the optional statement families can choose a binary that does not link their client and protocol dependencies. This narrows the shipped process's optional runtime surface. It does not reduce the repository's go.mod, source, CI, or dependency-maintenance surface, and it is not a security or vulnerability-free guarantee.

Size evidence

Built as a controlled relative comparison with the same Go toolchain and target, CGO_ENABLED=0, -trimpath, and -s -w. These are not claims of byte identity with GoReleaser archives:

Target	Full	Slim	Reduction
darwin/arm64	58,887,538 bytes	50,564,274 bytes	8,323,264 bytes (14.13%)
linux/amd64	60,424,352 bytes	51,798,176 bytes	8,626,176 bytes (14.28%)

Validation

• goreleaser check --config .goreleaser.yaml
• full and slim runtime surface checks
• version and installation-source ldflags checks for both binaries
• controlled full/slim builds for darwin/arm64 and linux/amd64
• go test -short ./...
• make check with golangci-lint v2.13.2
• git diff --check issue-778-goreleaser-scaffold...HEAD

Follow-ups

Registration-time mutation-policy enforcement and feature-package completion construction are separate seam-contract gaps and are not required to link the slim binary. They remain separate follow-ups rather than expanding this release-artifact change.

Depends on #789.

[github-actions]

Code Metrics Report

📊 View detailed coverage report (available for 7 days)

	main (b7c135d)	#858 (29d185e)	+/-
Coverage	74.0%	74.0%	-0.1%
Code to Test Ratio	1:1.3	1:1.3	+0.0
Test Execution Time	1m42s	1m43s	+1s

Details

  |                     | main (b7c135d) | #858 (29d185e) |  +/-  |
  |---------------------|----------------|----------------|-------|
- | Coverage            |          74.0% |          74.0% | -0.1% |
  |   Files             |             94 |             95 |    +1 |
  |   Lines             |           8093 |           8096 |    +3 |
  |   Covered           |           5993 |           5993 |     0 |
+ | Code to Test Ratio  |          1:1.3 |          1:1.3 |  +0.0 |
  |   Code              |          18876 |          18885 |    +9 |
+ |   Test              |          26133 |          26180 |   +47 |
- | Test Execution Time |          1m42s |          1m43s |   +1s |

Code coverage of files in pull request scope (88.5% → 87.5%)

Files	Coverage	+/-	Status
cmd/spanner-mycli-slim/main.go	50.0%	+50.0%	affected
internal/mycli/memefish_parser.go	76.9%	-23.1%	affected
internal/mycli/statement_processing.go	91.3%	+1.2%	affected
internal/mycli/statements_split_points.go	84.6%	-1.0%	affected
main.go	50.0%	+50.0%	modified

---

Reported by octocov