Pull popular crates from crates.io

[NobodyXu]

Currently we uses statistics from cargo-binstall (given that the quickinstall fetcher is not disabled) and cargo-quickinstall.

Users might still experience "install from source" before the quickinstall collects the stats and build them.

We should pull popular crates from crates.io and build them proactively for better user experience.

[NobodyXu]

There's a summary API which is used in crates-io-api that can be used to obtain most downloads, just updated, new crates, most recently downloaded crates.

https://docs.rs/crates_io_api/0.8.1/crates_io_api/struct.AsyncClient.html#method.summary

[NobodyXu]

I also think the cronjob needs to avoid curling crates.io index too much by switching to sparse index instead.

Edit:

The sparse index is much harder to parse in bash and doing it python would require additional packages to be installed.

[NobodyXu]

Turns out that the crates.io api only give 10 crates and they are mostly library crates.