bounds for timeout valye

[JarbasAl]

          If there’s a possibility the timeout can be changed at runtime to an unsuitable value, it might be worth the hit from setting it where it was before, but with bounds.

Otherwise, this solves the problem.

Originally posted by @ChanceNCounter in #58 (review)

from matrix chat

@JarbasAl

what should that upper limit be? this is how long a adapt context can remain valid
if we want it to be valid forever, lets support -1 for no timeout
otherwise what is sane? 1 day?

@ChanceNCounter

There’s probably an answer here, but idk how to estimate it. A day seems like overkill, but maybe not idk