It still works because... luck? The only thing that uses this is the
Crew Monitor which sends a Z-level as the argument... Which happens to
work because the Z-level is usually 1.
* Update NanoUI to the latest version, ported from Baystation12
* Update the Air Alarm UI
* Make changes to subsystems and some machines to make the new NanoUI
work
When the MC is making up missed subsystem fires from lag, it will now only fire that subsystem at most, half of that subsystem's normal fire rate until missed fires are made up, rather than firing as quick as possible making lag worst.
When a subsystem causes byond to miss a byond tick, it will stop processing subsystems and sleep for two extra byond ticks on top of its normal sleep rate.
When the cpu is above 80%, the MC will also sleep for twice as long between ticks (stacks with the change above)
The processing interval of the MC is now capped to a lower bound of two byond ticks or 1ds, whatever is lower.
Dwait now scales with the cpu usage var. 10% cpu adds 10% extra to dwait, 50% cpu adds 50% extra to dwait, etc.
Adds crew monitor to the asset cache system
Adds paper to the asset cache system
Added a way to send files to the client slowly without clogging up the queue. (This isn't technically "safe", but the client would only have issues if they didn't have that file already (rare), and only if they used a window that needed that asset (currently only nanoui windows) within the first 15 to 90 seconds of connecting (depending on ping))
Makes NanoUI use this slow send system to blindly send it's asset files. (Once bay's nanoui is ported, i'll improve this so that nanoui fully implements asset cache checking)
Makes asset cache use this system to send all registered asset files, so that ui windows don't have to wait for them if the client's been connected a while.
Replaces typesof(path) - path with subtypesof(path) in obvious places. I was a bit conservative, there's probably a few more places that could use this.
