Add --output <file> flag to write results to a file

[vmvarela]

Description

sql-pipe currently writes results only to stdout. In automation scripts or when building pipelines with multiple steps, users often want to write results directly to a file without shell redirection. A --output flag makes the intent explicit and avoids accidental overwrites common with >.

Example

# Without --output (works but fragile in scripts)
sql-pipe 'SELECT * FROM t' < input.csv > results.csv

# With --output (explicit, safer)
cat input.csv | sql-pipe --output results.csv 'SELECT * FROM t'

# Combining with --json
cat input.csv | sql-pipe --json --output report.json 'SELECT region, SUM(revenue) FROM t GROUP BY region'

Acceptance Criteria

• --output <path> writes stdout to the given file path, creating or overwriting it
• If the file cannot be created (bad path, permissions), exit with code 1 and a clear error message
• Without --output, behaviour is unchanged (write to stdout)
• Compatible with all output modes: CSV (default), --json, --header
• Documented in --help, README.md Flags table, and docs/sql-pipe.1.scd
• Test: output written to file, file contains correct content, missing parent directory error handled

Notes

• Implementation: open file before query execution, pass writer to execQuery
• Atomic write (write to temp file, rename) is a nice-to-have but not required for v1