Log actual recorded events on panic in `match_events!` (#4069)

## Motivation and Context
We've had flaky tests in our release pipeline. The pain points are that
reproducing these failures is difficult and that when a test does fail,
`panic` from the `match_events!` macro does not provide visibility into
the actual recorded events.

This PR does not address the underlying test failures but aims to
improve visibility into the actual recorded events when a test fails.
This will help us better diagnose and troubleshoot flaky tests.

Here is the new diagnostic message looks like (deliberately made a test
fail just to show what the message looks like):
```
  expected `ev!(connect)` but got Response(HttpResponse { status: 429, body: b"{\"__type\":\"com.amazonaws.dynamodb.v20120810#ThrottlingException\",\n\"message\":\"enhance your calm\"}" })
  actual recorded events: [DnsLookup("this-url-is-converted-to-localhost.com"), NewConnection, Response(HttpResponse { status: 429, body: b"{\"__type\":\"com.amazonaws.dynamodb.v20120810#ThrottlingException\",\n\"message\":\"enhance your calm\"}" }), Response(HttpResponse { status: 200, body: b"{\"Count\":0,\"Items\":[],\"ScannedCount\":2}" })]
```

----

_By submitting this pull request, I confirm that you can use, modify,
copy, and redistribute this contribution, under the terms of your
choice._