Improve auto CQ

[zcsahok]

Auto CQ came up in two issues recently (#463 and #466).

Let's see if we can improve here.

Stop auto CQ on callsign grab (from #463): normally the callsign field is empty during CQ, so check if it has been set by a grab (clicking on Fldigi RX window). In addition to key presses consider such events also as CQ-stopper, but don't report them to the caller.

[zcsahok]

Handle PgUp/Dn (keyer speed up/dn) internally (from #466): make use of handle_common_key() after filtering.
UI experience is not the best due to the delayed update, but one can live with it.

[df7cb]

I confirm fldigi grabbing stops autocq with this patch. Thanks!

[zcsahok]

Added handling of Ctrl/Alt-PgUp/Dn (CQ delay control), so #466 is fully covered IMO.

Due the specific nature of these key had to restructure key handling a bit: mapping to non-curses keys happens in `onechar()'.

[zcsahok]

Changing CQ delay has only an immediate effect when shortening it. Lengthening will be effective at the next CQ round. I think this is tolerable.

[dl1jbe]

Changing the keying speed results in CQ delay starting early or late depending on the direction of change.
'message_time' should be recalculated at every start of the while() loop. So both will get in sync again after the next round.

Better yet would be to recalculate the remaining time as soon as the CW speed got changed.

[zcsahok]

Moved message length calculation into the loop.

We have no definite info how much of the message is left when speed changes. It could be possible to estimate it, but it's not worth the effort IMO. For small speed changes (+/- 2 wpm) the difference is hard to notice.

[dl1jbe]

Moved message length calculation into the loop.

Thanks. Looks good now.

We have no definite info how much of the message is left when speed changes. It could be possible to estimate it,

Yes if first calculation is correct we could estimate based on the percentage of time already spent.

but it's not worth the effort IMO. For small speed changes (+/- 2 wpm) the difference is hard to notice.

True. It only matters for bigger changes.