Improve Monitoring Report Interval for low allowable_throughput_failure_interval_seconds

[waahm7]

Description of changes:
The report interval was 1 second by default, which can be slow if allowable_throughput_failure_interval_seconds is also low like 1 second in the IMDS Client. It took around 3 seconds to figure out if the connection was stuck. This change improves that time to around 1.5 seconds.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[graebm]

We talked offline and you gave the stats of:

/3 = ~1.6 seconds
/5 = ~1.4 seconds
/10 = ~1.2 seconds

maybe we should go for /10? I mean ... being 20% off still isn't fantastic, and if you're setting the timeout to 1sec, you obviously want that timeout to be fast

[TingDaoK]

Oh, we are using minimum_throughput_bytes_per_second.

So that if we are sampling the data less than 1 secs, we will have a inaccurate rate per second :(

The failing test has

    struct aws_http_connection_monitoring_options monitor_opt = {
        .allowable_throughput_failure_interval_seconds = 1,
        .minimum_throughput_bytes_per_second = 1000,
    };

The test failing because is doing a 900 Byte transfer after a second, so as we sampling the data every 300 ms, there will be two sampling result in 0 B/s, but one sampling results in 3000 B/s.

And the test allows 1 secs to failure. So, in 1 secs, there always be one sampling is larger than the minimal, so that it never shutdown the connection as expected.

:(