bug(android): bad order of operations for updating keyboard state

While collecting data via profiling for a different task, I ran into proof of an internal race condition:

[Chrome performance profile (.json)](https://github.com/keymanapp/keyman/files/14860632/sus.android.keyboard.swap.json)

Note the timestamps in the following images:


Web has been told to load a keyboard; this, of course, requires fetching it.

The usual _post-keyboard load_ update:


The keyboard script _actually_ loading, being processed, and moving toward display of the OSK:


Note: _**those last two steps are out of order!**_

Fortunately... we do a context-reset anyway once the keyboard load is complete, so I _believe_ that, at present, this has no net effect. (So, I've marked it as low-priority for now.) The fact that this can occur does appear significant though, so I wanted to note it in case it matters at some point in the future. I'd have assumed that this ordering would never occur.