bugfix in flush key at 2.0 branch

[randy3k]

1. only invalidate ui when some keys are processed
2. do not start_timeout if the last key is _Flush, otherwise it causes a dead loop
3. only send flush key when the queue is not empty

close #570

[codecov-io]

Codecov Report

Merging #576 into 2.0 will decrease coverage by 0.1%.
The diff coverage is 100%.

@@ Coverage Diff @@ ## 2.0 #576 +/- ## ========================================== - Coverage 71.08% 70.98% -0.11%  ========================================== Files 123 123 Lines 11885 11892 +7 ========================================== - Hits 8448 8441 -7  - Misses 3437 3451 +14

Impacted Files	Coverage Δ
prompt_toolkit/key_binding/key_processor.py	90% <100%> (-1.55%)	⬇️
prompt_toolkit/layout/dummy.py	60% <0%> (-33.34%)	⬇️
prompt_toolkit/application/current.py	85.18% <0%> (-7.41%)	⬇️
prompt_toolkit/application/dummy.py	68.75% <0%> (-6.25%)	⬇️
prompt_toolkit/layout/controls.py	74.76% <0%> (-0.32%)	⬇️
prompt_toolkit/application/application.py	77.41% <0%> (-0.27%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ea5e381...12edf61. Read the comment docs.

[jonathanslenders]

Thank you @randy3k!
This was a serious bug. I really appreciate it!
I'll review and merge tomorrow.

[randy3k]

@jonathanslenders I think that it should check key_buffer instead of input queue, right?

[jonathanslenders]

That's right. Thanks @randy3k
Fixed here: 0a64979