Kai Mast

We need to merge #3802 first, which measures BFT sync performance. This PR does not change P2P/client sync.

Looks good to me! I pushed one more commit with some minor cleanups ([83355c6](https://github.com/ProvableHQ/snarkOS/pull/4029/commits/83355c6e11a2f431e856c371922909b170a95b37)). I also rebased https://github.com/ProvableHQ/snarkOS/pull/3902 on it. That PR has some networking changes and the tests succeed....

The following case is problematic: Consider two clients (C1 and C2) and one validator V. C1 is connected only to C2, and V is connected to C2 (and other validators,...

> Given this is not closed, does it mean one cannot achieve headerbars akin to GTK's/Adwaita's in Iced? Just curious. You can disable window decorations in winit and then draw...

I added `read_le_with_unchecked`, because it made implementing the unchecked versions a lot cleaner. Otherwise, it requires a lot of if branching. > The only thing I see to improve is...

For the block sync work, I have been using tracing directly (see [this PR](https://github.com/ProvableHQ/snarkVM/pull/2887)). That might be another option if these calls happen too frequently to report them to Prometheus.

I have also been slowing doing this, for example, in #3050. For unrecoverable errors, I don't think it matters as much, though, as those will simply be logged or passed...

Good points. We should definitely aim to use `thiserror` for new API calls. I am also happy to review any changes you need to make to existing code. Victor probably...

In #3789 I am also adding serialization of `anyhow::Error` to the API (see [here](https://github.com/ProvableHQ/snarkOS/pull/3789/files#diff-fbe1fdc35fe9fe6a93da50c81e90f7f528a1a35dce4883a53942592a63cc54edR86)). A unified crate should include this functionality as well. It could rely on functionality introduced in...

I would prefer this too. Having `dev_println` enabled for `test_network` in snarkOS creates a lot of output that cannot be hidden. My guess is the reason for a dedicated `dev_println`...