feat(indexer): preflight endpoint, sync with abstract discovery provider interface

[m-kus]

TL;DR

Refactored the Discovery Service API with improved endpoint naming, enhanced cursor structure, and added a new preflight endpoint for transfer readiness checks.

What changed?

• Renamed /v1/discovery/incoming/sync to /v1/sync/incoming_state for consistency
• Added new /v1/discovery/preflight endpoint to check transfer readiness
• Improved cursor structure:
  • Changed total_n_subchannels to skip_subchannel_discovery boolean flag
  • Added skip_channel_discovery flag to optimize channel discovery
  • Added max_note_index to optimize note scanning with exponential probing
  • Keyed channels by sender address instead of channel key for better organization
• Added helper functions for computing channel and subchannel markers
• Reduced default max_reads from 1000 to 50 (capped at 100) for better request sizing
• Enhanced response structure with more consistent field naming and organization

How to test?

1. Use the new /v1/sync/incoming_state endpoint to discover incoming notes
2. Test the /v1/sync/outgoing_state endpoint to discover outgoing channels
3. Try the new /v1/discovery/preflight endpoint with:

   {
     "sender_address": "0x...",
     "viewing_key": "0x...",
     "recipient": "0x...",
     "token": "0x..."
   }

4. Verify the preflight response contains the expected flags:

   {
     "block_ref": "0x...",
     "sender_registered": true,
     "channel_exists": true,
     "subchannel_exists": true
   }

Why make this change?

These changes improve the API's consistency, efficiency, and usability:

1. The preflight endpoint enables clients to quickly check transfer readiness with minimal storage reads (max 4)
2. The improved cursor structure reduces unnecessary storage reads by skipping already-discovered channels/subchannels
3. The exponential probing for note discovery optimizes scanning for sparse note distributions
4. Consistent naming and organization makes the API more intuitive and easier to use
5. Smaller default read limits help prevent excessive resource usage while still allowing pagination

---

This change is 

[m-kus]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• feat(indexer): e2e discovery flow #440
• feat(indexer): preflight endpoint, sync with abstract discovery provider interface #439 👈 (View in Graphite)
• feat(indexer): outgoing state sync endpoint #438
• feat(indexer): batch state sync #435
• feat(indexer): sync outgoing state #433
• feat(indexer): discover outgoing channels #430
• feat(indexer): find last used note index #429
• feat(indexer): outgoing channel primitives #424
• feat(indexer): filter out spent notes, use secret felt #422
• feat: scaffold e2e testing harness, move devnet dump generation to a separate script #421
• feat(indexer): sync code and spec, update guidelines and todos #414
• feat(indexer): add incoming state sync endpoint #413
• feat(indexer): api error types #412
• refactor(indexer): common helpers for spawning devnet/indexer #411
• feat(indexer): incoming state sync orchestration #410
• feat(indexer): cursor pagination for discovery #408
• refactor(indexer): reorganize discovery-core, prepare for state sync #407 : 1 other dependent PR (#453 )
• feat(ci): speed up rust workflow #415
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.