Random performance improvements

[yawkat]

I have a very large quassel core with many buffers and many users in those buffers, and quasseldroid is very slow to connect (4 minutes) for me. These changes attempt to fix some of the most obvious issues, cutting down connect time by 50% or 2 minutes for me.

Some of these solutions are more patching up than real fixes (especially that Executor...) but it works.

Feel free to cherry-pick individual changes from this PR, or split it into multiple commits, I don't really care since it runs fine on my phone now :)

[justjanne]

This is actually very close to what I’ve done for the rewrite, but awesome!

I’ll merge this right now.

[jgkamat]

@justjanne I haven't tested this out, so I'm not sure if it works, but a faster connection time would be very nice to have, so if this works, can it be mered into master?

[yawkat]

These changes do work, but they are far from an ideal solution. They were hacked together in half an hour and this PR is only meant as a hint on important bottlenecks in the connection phase.

[justjanne]

@yawkat: There’s still an issue with the ObjectName – nextUniqueInteger won’t work, because the ObjectName is defined on the server.

If you change that part, the rest seems to work nicely, so I’d be willing to merge that.

[justjanne]

@yawkat Also, could you update the BuildTools and android-sdk versions? Then Travis CI should be able to build it, too (and the checks would work)