2.x: Fix window() with start/end selector not disposing/cancelling properly

[akarnokd]

This PR fixes the start-end selector variant of Observable.window and Flowable.window to properly

• dispose the window-opening sequence upon disposing the main output flow and
• disposing the main upstream upon disposing the main output flow provided there are no windows open.

Fixes: #6397

[codecov]

Codecov Report

Merging #6398 into 2.x will increase coverage by 0.01%.
The diff coverage is 88.88%.

@@             Coverage Diff              @@
##                2.x    #6398      +/-   ##
============================================
+ Coverage     98.24%   98.26%   +0.01%     
+ Complexity     6294     6292       -2     
============================================
  Files           675      675              
  Lines         45156    45162       +6     
  Branches       6239     6243       +4     
============================================
+ Hits          44365    44378      +13     
+ Misses          247      243       -4     
+ Partials        544      541       -3

Impacted Files	Coverage Δ	Complexity Δ
...ators/flowable/FlowableWindowBoundarySelector.java	90.96% <88.88%> (+4.17%)	2 <0> (ø)	⬇️
...s/observable/ObservableWindowBoundarySelector.java	91.56% <88.88%> (+0.76%)	2 <0> (ø)	⬇️
.../io/reactivex/internal/schedulers/IoScheduler.java	90.42% <0%> (-3.2%)	9% <0%> (ø)
...a/io/reactivex/internal/util/QueueDrainHelper.java	97.22% <0%> (-2.78%)	56% <0%> (-2%)
...rnal/operators/observable/ObservableSwitchMap.java	93.61% <0%> (-2.66%)	3% <0%> (ø)
...activex/internal/observers/QueueDrainObserver.java	97.43% <0%> (-2.57%)	21% <0%> (-1%)
...x/internal/operators/flowable/FlowablePublish.java	96.2% <0%> (-2.11%)	11% <0%> (ø)
...activex/internal/schedulers/ScheduledRunnable.java	98.07% <0%> (-1.93%)	29% <0%> (-1%)
...tivex/internal/operators/single/SingleTimeout.java	98.33% <0%> (-1.67%)	2% <0%> (ø)
...io/reactivex/subscribers/SerializedSubscriber.java	98.86% <0%> (-1.14%)	26% <0%> (-1%)
... and 23 more

---

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 6e266af...b2e8c2b. Read the comment docs.

[vanniktech]

Why on it's own line?

[akarnokd]

A habit if more methods are ever chained onto them.

[vanniktech]

Nice that we can fix this too!