Rename seconds to TTL and use std::chrono

[ljbade]

This PR is set to merge on the rebased master, but I will update that once #7 is merged.

I found the use of Seconds() confusing for the metric expiry time, so I renamed it TTL().

I also updated the expiry code use use std::chrono as that is already used in the Summary class, and makes the TTL units clearer.

Finally I added a bunch of test cases to check that the expiry stuff is working correctly.

[coveralls]

Pull Request Test Coverage Report for Build 2063488925

• 17 of 18 (94.44%) changed or added relevant lines in 7 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.2%) to 96.306%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
core/src/gauge.cc	4	5	80.0%

Totals
Change from base Build 2062151750:	0.2%
Covered Lines:	756
Relevant Lines:	785

---

💛 - Coveralls

[ljbade]

I will be needing the chrono stuff as well for PR I am currently working on where I need to feed in external time to the summary metric type.

[silverjam]

This look even better for an upstream PR submission, in general we should make reasonable effort to get things upstream so that our forks are as minimal as possible

[ljbade]

@silverjam yep I think this should be good for upstream. I think replacing steady clock with system clock might be controversial but that was the only way I could then pass in time stamps from recorded data in for replay. I did try to think if there was a way to do make it flexible with templates but ended up a bit of a mess.

[silverjam]

@silverjam yep I think this should be good for upstream. I think replacing steady clock with system clock might be controversial but that was the only way I could then pass in time stamps from recorded data in for replay. I did try to think if there was a way to do make it flexible with templates but ended up a bit of a mess.

Hrmm, OK, maybe we could upstream everything except that change then?

[ljbade]

Yeah I think that would make sense. I feel like us replaying data is probably a bit of a corner case for this library...