[core-amqp] Update retry to remove final sleep after all attempts are exhausted

[chradek]

Reference: #12587 (comment)

Currently the retry function in @azure/core-amqp contains a loop that includes a sleep that matches the user's retryDelayInMs if a retryable error is encountered while invoking the provided operation.

azure-sdk-for-js/sdk/core/core-amqp/src/retry.ts

Lines 232 to 244 in 1c2e8c4

	if (lastError && lastError.retryable) {
	logger.verbose(
	"[%s] Sleeping for %d milliseconds for '%s'.",
	config.connectionId,
	targetDelayInMs,
	config.operationType
	);
	await delay(
	targetDelayInMs,
	config.abortSignal,
	`The retry operation has been cancelled by the user.`
	);
	continue;

In the scenario where the final allowed attempt fails with a retryable error, this function will still apply the sleep before throwing the error. In @azure/event-hubs, which uses a default retryDelayInMs of 30000 (30 seconds), this means the user's application has to wait an extra 30 seconds between when the last error is encountered and when it is thrown to the user's code.

Instead, we should add a check to see if the total number of retry attempts have been exhausted, and if they have, skip the sleep.