Caching NetworkInferences

[xmariachi]

Purpose of Changes and their Description

• Stores Network Inferences and Outlier-Resistant Network Inferences
• Keeps them the same amount of time, pruning as usual after rewards are done.
• They get filled up at CloseWorkerNonce. A test suite has been added to worker_test.go(there were none) to test validity of the function in general.
• Related migration changes related (genesis, query)
• Removes duplicity of GetLatestNetworkAvailableInferences and GetLatestNetworkInferences. Keeping the latter name.
• Fix reward_test.go tests force-advancing to the end of the wsw.
• Removes coefficients, weights and also lossBlockHeight from the GetLatestNetworkInferencesResponse.

---

• Adds validations for input layer objects

Are these changes tested and documented?

• If tested, please describe how. If not, why tests are not needed. -- Current unit test fixed, new unit tests added.
• If documented, please describe where. If not, describe why docs are not needed. -- This is an internalization and fixing of issues, though these are not documented nor need further documentation.
• Added to Unreleased section of CHANGELOG.md?

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf Linter / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	⏩ skipped	Mar 27, 2025, 6:11 PM

[amimart]

Nice work, I've let a few remarks

[guilherme-brandao]

Just did a test adding max inferers and forecasters in the network inference generation and it took approx 300ms for just one topic. We need to discuss this number before moving forward since now we will be processing the network inference in the EndBlock.

[xmariachi]

Just did a test adding max inferers and forecasters in the network inference generation and it took approx 300ms for just one topic. We need to discuss this number before moving forward since now we will be processing the network inference in the EndBlock.

Different other efforts are being done separately to improve performance.

[amimart]

LGTM, just a minor remark up to you :)

[xmariachi]

Rebased to @guilherme-brandao 's work on #772.
Also found a problem while testing with offchain nodes: latest inferences are taking the last worker commit which is wrong. Should just get the latest from the store.

[amimart]

Looks good! 🚀