chore: more retries for initial setup client

[sgbalogh]

We use this repo internally within an Antithesis env, so network failures are expected. The actual log collection portion of this bin tolerates that (the errors become part of the log), but I found it was possible to crash during initial basin creation, before the actual test begins, if startup happens while network faults are occurring.

[greptile-apps]

Greptile Overview

Greptile Summary

Introduces separate retry configurations to prevent crashes during initial basin/stream creation in Antithesis environments with network faults.

• Created setup_config with aggressive retry settings (1024 max attempts, 1s min delay) for initial setup operations (basin/stream creation, initial tail check)
• Created test_config derived from setup_config but with default retry behavior and AppendRetryPolicy::NoSideEffects for actual test operations
• Removed duplicate S2::new(config.clone()) call after initial setup
• Added necessary imports: NonZeroU32 and Duration

The change correctly separates concerns: setup operations can tolerate long retry periods since they happen before the test begins, while test operations maintain the original retry behavior to ensure test accuracy.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• The change is well-structured, addresses a real operational issue (crashes during network faults in Antithesis), and correctly separates retry behavior between setup and test phases. The aggressive retry settings (1024 attempts) are appropriate for initial setup in fault-injection environments, while test operations maintain original behavior for accuracy.
• No files require special attention

Important Files Changed

Filename	Overview
rust/s2-verification/src/bin/collect-history.rs	Added aggressive retry configuration (1024 attempts, 1s delay) for initial setup to tolerate network faults in Antithesis environment, while keeping separate config for test operations

_{Last reviewed commit: 992e7dd}