BlockSync provider

[snaumov]

Summary of changes
Changes introduced in this pull request:

BlockSync "provider" - the logic to run for an incoming BlockSync libp2p request.

Test notes

It was somewhat hard to test this MR. I've added a unit test prepopulating MemoryDB with some chain data (thanks @austinabell for generating file for me), and compared the response with FilFox, see comments in the test.

Reference issue to close (if applicable)

Closes #673

Other information and links

[austinabell]

Thanks so much for this contribution! Looks generally good, and happy to help with any of the changes or suggestions if you would like :)

[austinabell]

These can be bit operations, BLOCKS is the least significant bit, and MESSAGES is the second least significant bit.

Can be something like:

    pub fn include_blocks(&self) -> bool {
        self.options & BLOCKS != 0
    }

    /// If a request expects messages to be included in response.
    pub fn include_messages(&self) -> bool {
        self.options & MESSAGES != 0
    }

[snaumov]

feels a little less readable imo 🤔

[austinabell]

Don't need this

[snaumov]

it's in a spec though?
https://spec.filecoin.io/#algorithms__block_sync

[austinabell]

yeah it's not necessary though, that only exists because it wasn't bit flags before. 0b01 | 0b10 == 3. There is no need to use this constant itself.

Edit: Also that blocksync spec is wildly incorrect, don't use it for anything meaningful

[snaumov]

yeah it's not necessary though, that only exists because it wasn't bit flags before

Sorry, when you say before are you referring to your suggestion here: #724 (comment), or these are bit flags in a spec now too?

[austinabell]

LGTM, thanks so much for the contribution. The only comments left are opinionated and I'm good with this :)

[austinabell]

Sorry was just waiting for branch to update before pulling in, didn't want to merge into your branch if you were making changes. I just updated the branch and will pull in when CI passes. Thanks!