refactor!: dropping pagination from getLogsByTags

[benesjan]

In this PR I drop pagination from getLogsByTags endpoint because pagination here doesn't really make sense because we get more logs per tag only if there are multiple devices sending logs from a given sender to a recipient.

This is the final of 3 PRs in which I clean up the getLogsByTags endpoint:

1. First PR - including block timestamp in return value,
2. This is the second PR - I type the tag arg of the function to be SiloedTag,
3. in the last PR I will drop pagination from this endpoint.

Potential DoS vector

Log tag is unconstrained and hence can be arbitrarily chosen by the sender. This introduces a potential DoS vector because the node currently loads all the logs into memory and returns them. Hence it could be feasible to re-use the same tag thousands of time and then spam a node by request for this tag.

I think the solution here is to simply define the maximum number of logs a node is willing to store per tag in the db. Given that there is no legitimate use case for log reuse I think making it even as low as 5 would be fine.

Note that this PR didn't introduce this DoS vector because even though I dropped the pagination here it didn't seem to be enforced before (if limitPerTag arg was undefined we just returned it all). For this reason I think we can merge it and tackle the issue in a followup PR.

[benesjan]

• refactor!: expanding return value of getLogsByTags #19233
• feat: integrating new log sync #19125
• refactor!: dropping pagination from getLogsByTags #19161 👈 (View in Graphite)
• refactor!: splitting getLogsByTags to private and public versions #19158
• refactor!: optimizing AztecNode::getLogsByTags for new log sync algo #19157
• refactor: more robust tagging index sync as recipient #19030
• refactor: more robust tagging index sync as sender #17611
• refactor: renaming IndexedTaggingSecret as PreTag #17568
• refactor: returning last used tagging index #17534
• feat: ExecutionTaggingIndexCache #17445
• refactor: tagging related cleanup #17451
• next

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

[spalladino]

Note that this PR didn't introduce this DoS vector because even though I dropped the pagination here it didn't seem to be enforced before (if limitPerTag arg was undefined we just returned it all). For this reason I think we can merge it and tackle the issue in a followup PR.

Heads up that's not correct. The limit was enforced at the schema level (maybe not the wisest decision in retrospective, since it's unclear from the code), where we'd default the limitPerTag to the max value if not set. Still, I'm fine if you want to just remove it, as long as you add some guardrails in a future PR.