Add --fail-on-unfixable flag for single-pass fix + validation

Problem

When using swiftlint --fix, the command always returns exit code 0, even when unfixable violations remain. This forces users to run SwiftLint twice in CI/pre-commit hooks:

swiftlint --fix && swiftlint --strict

This doubles the execution time for files with violations.

Proposed Solution

Add a new flag --fail-on-unfixable (or --strict-fix) that:

1. Performs auto-correction as usual
2. Returns non-zero exit code if any violations couldn't be auto-fixed
3. Optionally outputs the unfixable violations to stderr

Use Case

Pre-commit hooks and CI pipelines that want to:

• Auto-fix what can be fixed
• Block commits/builds when manual intervention is required
• Do this in a single SwiftLint invocation

Current Workaround

Using swiftlint --fix && swiftlint --strict, but this is suboptimal for large codebases and adds unnecessary latency to developer workflows.

Example Usage

# Current (two passes)
swiftlint --fix --config .swiftlint.yml file.swift
swiftlint --strict --config .swiftlint.yml file.swift

# Proposed (single pass)
swiftlint --fix --fail-on-unfixable --config .swiftlint.yml file.swift

Related Issues

• #5387 - Discussion about --strict --fix behavior

Start at the SwiftLint command-line handling for --fix and compare it with the behavior discussed in related issue #5387. Verify the proposed single-pass invocation against the current two-pass workaround, including the exit status and any output for remaining violations. Done means auto-fixable violations are corrected and unfixable violations produce a non-zero result.