feat: implement batch_create_streams for atomic multi-stream creation (closes #1450)

[JinadJay]

Summary

Implements batch_create_streams for issue #1450 — creating multiple streams atomically in a single transaction for gas efficiency.

Changes

• batch_create_streams(env, sender, params) public entrypoint in lib.rs:
  • Authenticates the sender exactly once
  • Validates every parameter (curve type, amount, time range, restricted addresses, combined-total overflow) before any state mutation — all-or-nothing
  • Enforces a batch size limit of 20 streams (MAX_BATCH_SIZE), returning Error::BatchSizeExceeded beyond it
  • Allocates sequential stream ids and persists all streams + user profiles in a single storage pass (single NEXTID read/write)
  • Returns the new stream ids in the same order as params
• StreamParams contracttype struct (receiver, token, total_amount, start_time, end_time, curve_type, is_soulbound)
• Error::BatchSizeExceeded variant
• 8 new tests in src/batch_test.rs: batch of 2, batch of 10, mixed parameters, partial-failure rollback, total overflow rollback, empty batch, max-size exceeded, restricted receiver rollback

Fixes required to make the branch compile & pass tests

The contributing branch was in a broken state after a bad merge — this PR also repairs it so CI can pass:

• Restored missing String import and missing Error variants (StreamEnded, MetadataLabelTooLong, TooManyTags, TagTooLong) referenced by existing code
• Fixed batch_withdraw's missing closing brace (syntax error)
• Moved stream metadata into its own persistent METADATA map (Option<StreamMetadata> inside the Stream contracttype is not serializable)
• Added the sender-ownership check to cancel_stream (matches the documented "only the sender may cancel" behavior and the existing security test)
• Fixed stress_test.rs create_many to stream to a single receiver and corrected the test_pause_resume_restores_vesting expectation in security_test.rs

Verification

• cargo check clean
• cargo test -- --test-threads=1 — all 70 non-stress tests pass (incl. 8 new batch tests); stress tests verified individually (they're slow due to the pre-existing O(n²) storage pattern in create_stream)

[Folex1275]

@JinadJay please resolve conflicts