#240 summarise: timeout/hang on repos with many workflow runs

Feature Request

swamp summarise --since 1d --json hangs/times out on repos with thousands of workflow runs. In our testing repo with ~3000+ workflow runs, the command never returns even after 30+ seconds.

Context

When using swamp at scale (e.g., running workflows in automation loops), the data volume grows quickly:

• 1.7GB in .swamp/ directory
• 1.1GB in data alone
• Thousands of workflow runs

swamp summarise appears to scan all runs within the time window without pagination or streaming.

Proposed Solutions

1. Add --limit N — cap the number of items processed/reported
2. Stream output — emit results incrementally as they're computed (especially useful for --json mode where an agent can start processing partial results)
3. Show progress — in interactive mode, show a progress indicator for large scans
4. Sample mode — --sample 100 to randomly sample N runs for statistical summary rather than exhaustive scan

Workaround

Use targeted queries instead:

# This works even with large datasets:
swamp data query 'modelName == "my-model"' --limit 10 --json
swamp workflow runs my-workflow --json | head -20

Environment

• swamp version: 20260504.233645.0-sha.430c1535
• Repo data size: 1.7GB, ~3000+ workflow runs