Add batch operations support

[zzstoatzz]

Problem

pdsx currently operates on single records at a time. For bulk operations, users must script loops around individual commands.

Proposed Enhancement

Add batch operation support for common multi-record operations:

Batch Delete

# Delete multiple posts by rkey
pdsx batch delete app.bsky.feed.post abc123 def456 ghi789

# Delete from file (one rkey per line)
pdsx batch delete app.bsky.feed.post --from-file ./to-delete.txt

# Delete with pattern matching
pdsx batch delete app.bsky.feed.post --filter 'text contains "spam"'

Batch Create

# Create multiple records from JSON array
pdsx batch create app.bsky.feed.post --from-file ./posts.json

# Create from JSONL (one record per line)
pdsx batch create app.bsky.feed.like --from-jsonl ./likes.jsonl

Batch Export

# Export all records from collection to file
pdsx batch export app.bsky.feed.post --output ./backup.json

# Export with pagination (all pages)
pdsx batch export app.bsky.feed.post --all --output ./all-posts.json

Batch Import

# Import records from backup
pdsx batch import app.bsky.feed.post --from-file ./backup.json

# Import with dry-run
pdsx batch import app.bsky.feed.post --from-file ./backup.json --dry-run

Implementation Notes

• Use com.atproto.repo.applyWrites for atomic batch operations
• Support both JSON array and JSONL formats
• Add progress bars for large batches
• Include rollback on partial failures
• Rate limiting to avoid API throttling

Benefits

• Efficient bulk operations
• Atomic multi-record changes
• Better for backup/restore workflows
• Reduces API round-trips

Priority

Low - nice to have, but current single-record operations work fine for most use cases

Related

• Could use existing pagination cursor support
• Complements high-level commands (Add high-level convenience commands #34)