Change optimize compaction holdoff start message to debug level

[davidby-influx]

This message is too noisy at info

[cpinflux]

It's certainly noisy, but I feel that it's currently being useful to point towards a range of things that otherwise have no logging around them at all.

I scraped papertrail for all occurrences of this message and ended up with this monstrosity:

For example here, cluster 3379f17d logged it over 3 million times in 2 hours. Each shard logging it every few seconds. In this instance though, there didn't seem to have an impact other than increasing the query response time by 250ms for that time period. There's clearly significant context missing for me to understand it, but if this holdoff timer is for an hour on each shard, what was causing that function to be called such a vast number of times?

If this is just going to be tuned down, are there other broader info level messages that could be added to replace it? Whilst I've no belief that this particular message is important, it does appear to be helpful with a number of issues which feel like they are 1.12 related where there are often absolutely no other logs of note at all.

[davidby-influx]

@devanbenz - why would the hold-off be called more than once/hour/shard?