PublishNew was allowing 1 new crate an hour rather than every 10 minutes

[carols10cents]

In #6875, @pietroalbini said:

With this PR, the rate limits are:

Action	Burst amount	Refill time
Publish new crates	5	1 every 10 minutes
Publish updates to existing crates	30	1 every minute
Yanking or unyanking	100	1 every minute

but the code didn't match that intention, which explains why we were getting so many emails asking for rate limit increases since that PR went out.

The problem was that the default_rate_seconds for PublishNew was set to 60 * 60, aka 1 hour, when the intention was that it would be set to 10 * 60, or 10 minutes.

I added a new test, and for PublishNew it fails with:

---- rate_limiter::tests::default_rate_limits stdout ----
thread 'rate_limiter::tests::default_rate_limits' panicked at 'assertion failed: `(left == right)`
  left: `[2023-08-11T19:15:38, 2023-08-11T19:25:38, 2023-08-11T19:35:38, 2023-08-11T19:45:38, 2023-08-11T19:55:38, 2023-08-11T20:05:38, 2023-08-11T20:15:38, 2023-08-11T20:25:38, 2023-08-11T20:35:38, 2023-08-11T20:45:38]`,
 right: `[2023-08-11T19:05:38, 2023-08-11T19:05:38, 2023-08-11T19:05:38, 2023-08-11T19:05:38, 2023-08-11T19:05:38, 2023-08-11T20:05:38, 2023-08-11T20:05:38, 2023-08-11T20:05:38, 2023-08-11T20:05:38, 2023-08-11T20:05:38]`', src/rate_limiter.rs:212:9

because the last_refill times get set to the time at which you're publishing if you're not rate limited.

The CI run for the first commit shows the failure, then the second commit shows the test passing (modulo the flaky test).

[pietroalbini]

Whelp, good catch!

[LawnGnome]

Thanks!