This reverts commit 5e2881ce.

Most math and backend storage was floats anyways, so most of these changes just half the data copied during function calls without reducing precision.

Otherwise MANUAL_CONTROL messages are still sent at the regular 50Hz. This reduces transmission bandwidth from 0.9kbps when the joystick is used continuously to 0.1kbps when it's idle.
To prevent possible issues from the last high-bandwidth message being lost and the system then not responding for .1s, the last messsage is double-transmit which should prevent those issues.

Replaced with only 1 output message with the result of the first serial port scan.

In the logplayer, the data stream is read ahead for messages, which is used to determine the size of the output bytearrays. Previously, this wasn't done correctly resulting in non-ideal behavior. Now, the next message is scanned properly, so the only bytearrays that are emitted from QGCMAVLinkLogPlayer are MAVLink messages in their entirety. This also has the benefit of skipping invalid/corrupted messages in the log that's being played back.

Because the indexes into the stackedwidget are all wrong

There were two separate actions for the simulation view, leading to
strange behaviour when it was selected.

Be more forgiving about component IDs, but warn developers. Do *not* translate developer debug messages (they would show up in the translation tool). Anyone without sufficient command of the english language will have a hard time contributing or debugging this application and we should not entangle the poor translators in this.

Clears the persisted settings for the QGC

This disables Windows C Runtime assert dialogs from popping up.