The motivation here is to address an issue with WritableStream instances wrapping a WritableStreamSink implementation not processing abort() calls proactively. Without the flag the abort() will be handled lazily, meaning that a consumer must be actively consuming the data written to the stream in order to trigger the abort and close the stream. This is problematic if the consumer has dropped or otherwise stopped reading from the writable's internal buffer ... the pending abort will never be processed and resolved. This means that pending writes and the pending abort itself will never be resolved in those cases. With the flag, the queue of pending writes is drained immediately when abort() is called and the stream is transitioned to an error state immediately. This should cause the underlying WritableStreamSink to be aborted and dropped immediately and the pending write and abort promises should all get settled.

internal ticket: EW-8607