More accurate emulation speed management

[adriweb]

As discussed between @jacobly0 and @runer112.
Basically about frame drops, being too early/late, catching up, etc.

09:01:46 <@runer112> if you're ahead, sleep until next_time
09:01:57 <@runer112> if you're less than 0.1s behind, full speed ahead
09:02:15 <@jacobly> should I go with 100Hz?
09:02:25 <@runer112> if you're more than 0.1s behind, manually pull up cur_time (or pull back next_time?) so the gap is only 0.1s, and then full speed ahead
09:02:52 <@runer112> err
09:02:59 <@jacobly> oh, so only drop a multiple of .1s at a time?
09:03:05 <@runer112> no
09:03:10 <@runer112> drop any excess over 0.1s
09:05:12 <@jacobly> but then we stay stuck at .1s behind?
09:05:33 <@runer112> you shouldn't
09:05:39 <@runer112> if it's behind, it should run unthrottled
09:05:46 <@runer112> it should be able to catch back up
...
10:39:24 <@jacobly> basically the question is should time spent in the debugger be dropped or should we just reset afterwards
10:41:52 <@runer112> I guess ideally, after resuming from something like debugging, you'd set the defecit to 0

[mateoconlechuga]

I dunno but I probably broke things. I'm not sure where last_time goes.

[mateoconlechuga]

bump

[mateoconlechuga]

Well, I broke things but everything still works so I guess it's probably fine then?

[runer112]

Mateo has suggested that this outstanding issue is the cause of the bug that can lock up CEmu for a while upon giving it focus if "Pause emulation on focus change" is enabled.

[adriweb]

This may take a bit more time to analyze and take care of, so we might as well postpone it for a later release, instead of making it block v1.1.
Will be mentioned as a caveat in the release note...

[adriweb]

Setting this as a release blocker and v1.1 again, following a discussion with @runer112:

<Runer112> my [release] veto has been constantly applied because of that bug
<Runer112> it has pretty negative usability implications
<Runer112> probably worse than the sum of all bugs in 1.0, from the perspective of the average user anyway
<Runer112> and from our perspective because everybody who downloads it will be reporting bugs about freezes and stuff

[jacobly0]

The GUI lock up is unrelated to this issue.

[adriweb]

With the changes from #304, what do we do about this issue?

[mateoconlechuga]

I wouldn't really call this a "bug" anymore.

[adriweb]

Closing this thanks to all the improvements since then, and probably especially the recent ones by @calc84maniac.