Also handle errors in try_again()

[hadley]

And improve docs to correctly calculate failure probabiltiies

[claude]

Claude finished @hadley's task —— View job

---

Code Review

• Read and analyze all changed files
• Review error handling logic
• Check documentation accuracy
• Provide feedback

Issues found:

1. Documentation calculation error in R/try-again.R:20,23: The failure probability calculations are wrong. For a 10% failure rate function:

   • try_again(1, ...) gives 10% × 10% = 1% failure (not 1% as claimed)
   • try_again(2, ...) gives 10% × 10% × 10% = 0.1% failure (not 0.1% as claimed)

   Should be try_again(times - 1, ...) for the documented probabilities.

2. Logic improvement: Error handling now properly distinguishes expectation failures from other errors, which is good.