async-flow: deeper wake testing

[erights]

At #9097 (comment) @mhofman writes:

Technically the flow woke up on its own (wakeAll in prepareAsyncFlowTools) for both the following reasons:

• it was defined as an eager waker (default)
• it was in a failed state from a bad replay in the previous test.

A different approach may be to split this testGoodReplay test in 2, have the first one redefine the flow using startEager = false so it gets removed from the eager wakers, and only resolve v3 in the next test to verify that resolving the vow actually wakes up the flow. We should also verify then that the flow wasn't awoken before the vow resolution (probably can set a variable right after resolution of v3, and check that the variable is set inside the guest flow)

Given "A different approach may be" and due to limited time and immediate demand, I'm filing this rather than addressing it in #9097 itself. Should still be addressed in a later PR.