feat(cli): support per-run LLM provider and model overrides

[Makishima]

Description

Implements #680 by adding per-run LLM provider and model selection for both review and scan.

• Adds the --provider flag alongside the existing --model flag.
• Supports providers configured under both providers and custom_providers.
• Applies strict configuration precedence: CLI flags > environment variables > config file.
• Prevents partial configuration strategies from being mixed.
• Preserves environment credential fallback for the selected built-in provider.
• Prevents a model configured for one provider from leaking into another provider selection.
• Adds the resolved provider and model to JSON output through the llm field.
• Documents the new flags, precedence rules, and JSON output in all localized READMEs.

Compatibility note: a complete OCR_LLM_* environment configuration now takes precedence over file-based provider configuration.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Refactoring (no functional changes)
• Documentation update
• CI / Build / Tooling

How Has This Been Tested?

• make test passes locally
• Manual testing:
  • Verified review --help exposes --provider and --model.
  • Verified scan --help exposes --provider and --model.

Additional verification:

• go vet ./...
• make build
• Focused resolver regression tests for cross-provider model selection
• Git diff formatting and scope checks

Checklist

• My code follows the project's coding style (go fmt, go vet)
• I have performed a self-review of my code
• I have added tests that prove my fix is effective or my feature works
• New and existing unit tests pass locally with my changes
• I have updated the documentation accordingly (if applicable)
• I have signed the CLA

Related Issues

Closes #680

[CLAassistant]

All committers have signed the CLA.

[github-actions]

✅ OpenCodeReview: Review complete: 0 finding(s) across 6 selected item(s).

[lizhengfeng101]

It is not recommended to add too much content to the README; instead, it is suggested to add the descriptions of these commands to the official documentation section.

ref：pages/src/pages/DocsPage.tsx

[Makishima]

Addressed in 1afd06e: the per-run provider/model documentation was removed from the localized READMEs and moved to the official localized CLI reference pages rendered by DocsPage.

One clarification regarding the added command-summary row: ocr scan and its ocr s alias are not introduced by this PR. They already exist in cmd/opencodereview/scan_cmd.go:

• Use: "scan [flags]"
• Aliases: []string{"s"}

The row only documents this existing command because the PR adds --provider and --model support to both review and scan. If you prefer keeping the documentation diff strictly limited to the new flags, I can remove the command-summary row.

It is not recommended to add too much content to the README; instead, it is suggested to add the descriptions of these commands to the official documentation section.

ref：pages/src/pages/DocsPage.tsx

[lizhengfeng101]

Thanks for this PR! The feature design is solid, with thorough tests and documentation.

One change needed: please keep the strategy resolution order unchanged for the non---provider path (i.e., maintain config → env → Claude Code env → Shell RC).

Reasons:

• The --provider short-circuit logic is independent of the strategy list order — the feature works either way
• The precedence change is a separate breaking change that would silently alter behavior for existing users
• There is no need to take on this compatibility risk

Specifically, the strategies slice in ResolveEndpointWithOptions should be restored to:

strategies := []struct {
    name string
    fn   func() (ResolvedEndpoint, bool, error)
}{
    {"OCR config file", func() (ResolvedEndpoint, bool, error) { return tryOCRConfig(configPath, opts) }},
    {"OCR environment", func() (ResolvedEndpoint, bool, error) { return tryOCREnv(opts.Model) }},
    {"Claude Code environment", func() (ResolvedEndpoint, bool, error) { return tryCCEnv(opts.Model) }},
    {"Shell rc file", func() (ResolvedEndpoint, bool, error) { return tryShellRC(opts.Model) }},
}

The corresponding tests (e.g., TestResolveEndpoint_OCREnvironmentPrecedesConfig) should be adjusted accordingly. Everything else LGTM.

[lizhengfeng101]

@Makishima please keep the strategy resolution order unchanged for the non---provider path

[Makishima]

Addressed in 67a0b14. Restored the legacy non---provider resolution order:

1. OCR config file
2. OCR environment
3. Claude Code environment
4. Shell rc file

The explicit --provider path still short-circuits before the strategy list, so per-run provider selection is unaffected. I also updated the precedence/fallback tests and all three localized CLI reference pages to match the restored behavior.

@Makishima please keep the strategy resolution order unchanged for the non---provider path