refactor(p2p): minor pre-IPC refactors [part 5/11]

[glevco]

Depends on #1164

Motivation

In preparation for multiprocess P2P, some method calls between HathorProtocol (which represents a single connection) and the rest of the full node (specifically, the ConnectionsManager, TransactionStorage, and some other indirect calls) will be transformed to IPC, that is, remote calls.

To make this possible, this PR changes some hardcoded calls to properties and such to method calls. This will make it easier to convert them to IPC in future PRs.

Acceptance Criteria

• Move get_default_capabilities() from HathorManager to HathorSettings and make HathorManager.capabilities non-optional.
• Implement shortcut methods on ConnectionsManager, HathorProtocol, TransactionStorage, and update usages accordingly.
• No changes in behavior.

Checklist

• If you are requesting a merge into master, confirm this code is production-ready and can be included in future releases as soon as it gets merged

[codecov]

Codecov Report

Attention: Patch coverage is 95.52239% with 3 lines in your changes missing coverage. Please review.

Project coverage is 84.95%. Comparing base (2bcd22c) to head (b9b84ea).

Files with missing lines	Patch %	Lines
hathor/p2p/manager.py	88.23%	2 Missing ⚠️
hathor/p2p/sync_v2/agent.py	80.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@                             Coverage Diff                             @@
##           refactor/remove-vertex-storage-protocol    #1165      +/-   ##
===========================================================================
- Coverage                                    85.04%   84.95%   -0.10%     
===========================================================================
  Files                                          326      326              
  Lines                                        25123    25150      +27     
  Branches                                      3836     3835       -1     
===========================================================================
  Hits                                         21365    21365              
- Misses                                        3034     3054      +20     
- Partials                                       724      731       +7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[msbrogli]

It seems better to simply pass an rng to the protocol. Isn't it?

[glevco]

The RNG should be the same in the ConnectionsManager and in HathorProtocol, right? At least for simulation purposes.

If that's the case, we can't simply pass the RNG between them because now the protocol will live in a different process. Otherwise, if we can have different RNGs, then yes, it would be simpler to just have its own RNG here.

[jansegre]

We could use a "derived" rng instance, where rng in the main process just generates a seed that is used to instantiate the rng in the subprocess.

There's really only one reason for using an rng instance instead of a global rng, and that is using seeds to reproduce test cases. It's an important for reproducing rare cases that happen randomly in tests.

Considering there's parallelism, some events can happen in different order regardless of the initial seed, and this will affect how we generate random bytes regardless if we query the main thread for random bytes or generate them locally, so in that sense I think generating them locally is better because it avoids an IPC call. And if we really need to we can focus on avoiding that parallelism in tests or simulating it in a reproducible way.

[msbrogli]

Why did you remove this? It seems unrelated to the PR and it would change the behavior of sync-v2.

[glevco]

This variable is unused in this context. It's just a local variable that we set here, and the method returns just below.

In fact, in the latest commit (b9b84ea) I completely removed as it's just used to toggle a debug message that would never change.