Fix panic in shouldRetry type assertion and remove deprecated rand.Seed

[Nepomuk5665]

Summary

• Fix potential panic in shouldRetry function by using comma-ok idiom for type assertion
• Remove deprecated rand.Seed call that was re-seeding on every backoff attempt
• Add test coverage for shouldRetry with various error types

Bug Description

Type Assertion Panic

The shouldRetry function in retries/retries.go used a direct type assertion:

e := err.(*Err)  // This panics if err is not *Err type

While this function is internal and typically receives *Err values from Wait and Poll functions, a direct type assertion is unsafe. If any code path passes a non-*Err error, the program will panic with a runtime error.

The fix uses the comma-ok idiom to safely handle any error type:

e, ok := err.(*Err)
if !ok || e == nil {
    return false
}

Deprecated rand.Seed

The backoff function called rand.Seed(time.Now().UnixNano()) on every invocation. This has two issues:

1. Deprecated since Go 1.20: The global random number generator is now automatically seeded
2. Wasteful: Re-seeding on every retry attempt is unnecessary and has performance implications

Testing

• All existing tests pass
• Added new test TestShouldRetryWithNonErrType that verifies the fix handles various error types without panicking

[github-actions]

If integration tests don't run automatically, an authorized user can run them manually by following the instructions below:

Trigger:
go/deco-tests-run/sdk-go

Inputs:

• PR number: 1438
• Commit SHA: 0cefa2609c108943d86c68300e2cce5ca343f817

Checks will be approved automatically on success.

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed in 15 days if no further activity occurs. If this PR is still relevant, please leave a comment or push new changes to keep it open. Thank you for your contributions.

[renaudhartert-db]

Hi @Nepomuk5665, thanks for taking the time to contribute.

I had a closer look at the shouldRetry change. I don't think the panic can actually happen — shouldRetry is a private function only called from Wait() and Poll(), and both take callbacks that return *Err. The type system guarantees that only *Err or nil ever reaches shouldRetry, so the direct type assertion is safe by construction. The new !ok branch is effectively a dead code path.

The rand.Seed removal on the other hand is a good cleanup.

Would you be open to scoping this PR down to the rand.Seed removal only, as an Internal Change? That part we can merge quickly.