Fix bug with increment read window

[graebm]

Issue:
Turns out, if aws_channel_slot_increment_read_window() was called while the "window update task" was running, the "window update task" would not get rescheduled to run again.

I was writing read-window-update tests for the python WebSocket bindings, and the test was failing as the read window approached 0. Here's how it played out:

• the WebSocket started with a read window of 500 bytes
• a 1000 byte payload (total frame size 1004) was sent to the WebSocket
• the HTTP handler sent along the first 500/1004 bytes
• the WebSocket processes this, but since the first 4 bytes were frame metadata that it doesn't count against the read window, it increments its read window by 4 bytes
• but due to this bug, the "window update task" DOES NOT get rescheduled
• and my test would fail because it received 4 less bytes than it expected

Description of changes:
Fix rescheduling logic for "window update task"

Rant about tests:
I lost a few hours trying to add a regression test here in aws-c-io. First, I tried adding to the existing socket_handler_echo_and_backpressure test to reproduce my issue, but it didn't fail! Turns out, the bug can only occur when you have 3+ handlers in the channel. Then I started modifying the read_write_test_handler so that it could serve as a proper "midchannel" handler that buffers up frames if necessary before sending them downstream. Then realized this would be more work complex than making a new type of test-handler that solely functions as a midchannel handler. Then realized at this point we'd be better off adding true integration tests for websocket that exercise the REAL stack of handlers, instead of a bunch of test handlers. Then realized this would take days, and I already had the new python tests I was working on that exercise the real stack of handlers... so yeah. I'll just rely on the tests I'm adding in aws-crt-python for now, but we should probably add integration tests in aws-c-http someday.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[sbSteveK]

solid name change