feat: Slurp mode and multi-document support

[daavidrgz]

Summary

Add support for processing NDJSON (newline-delimited JSON) streams and multi-document YAML files.

Proposed Behavior

NDJSON / JSON Lines (--slurp / -s)

# Process each line independently (default streaming mode)
echo '{"name":"alice"}
{"name":"bob"}' | gq '{ name }'

# Slurp all lines into an array, then query
echo '{"name":"alice"}
{"name":"bob"}' | gq -s '{ name }'
# Output: [{"name":"alice"}, {"name":"bob"}]

Multi-document YAML

# Process multi-document YAML (--- separated)
kubectl get pods -o yaml | gq '{ items { metadata { name, namespace } } }'

Streaming mode

# Process a continuous stream without buffering everything into memory
tail -f log.json | gq '{ level, message }'

Motivation

NDJSON is ubiquitous in real-world pipelines — application logs, API pagination, kubectl output, database exports, and event streams all commonly use this format. This is table-stakes for CLI data tools. Without it, users can't use gq in many of the most common data processing scenarios.

Considerations

• Should streaming be the default for NDJSON, with --slurp to collect into an array?
• How does multi-document YAML interact with the output format? (one doc per result vs. merged)
• Memory implications: streaming mode should process one document at a time without buffering
• Error handling: should one malformed line/document abort the whole stream or skip and continue?