feat: add --sample <n> flag for quick data preview with schema

[vmvarela]

Summary

• Adds --sample [<n>] flag that prints a #-prefixed schema block to stderr and the first n CSV rows (default 10) to stdout — combining --columns --verbose + SELECT * FROM t LIMIT n into a single invocation
• Schema block lists each column name and its inferred SQLite type (INTEGER, REAL, TEXT), aligned for readability
• Implies --header; compatible with --delimiter / --tsv; mutually exclusive with --json, --columns, --validate, and a query argument
• Type inference buffers max(100, n) rows before emitting output; exits after printing n rows without reading the full input
• Documented in --help, README.md, and docs/sql-pipe.1.scd

Example

$ cat sales.csv | sql-pipe --sample 3
# Schema (3 columns):
#   id      INTEGER
#   region  TEXT
#   amount  REAL
id,region,amount
1,North,1250.00
2,South,875.50
3,East,2100.75

Closes #89