fix: Adding stack trace to panic catch errors

[deregtd]

Pond automatically catches panics without any option to surface them, but the wrapped error doesn't contain stack information so it's nearly impossible to track down the issue as is. This adds a stack trace to the wrapped error generated out of the recover function.

[alitto]

Hey @deregtd!
The best way to obtain the panic thrown by a task is to call the Wait() method on the task object returned by the call to pool.Submit(), these are the docs. If the task panics instead of returning an error, the panic is caught and returned wrapped in pond.ErrPanic. This way the error preserves the stack trace.

[deregtd]

Forgive my naivete here, but I tried for quite a while to get the stack trace out of the returned error from that, and couldn't find it stored anywhere in the wrapped error, eventually leading me down the road to making this fork to be able to get the stack trace. Other posts like this confirmed my findings. But I could easily be missing something.

[alitto]

You are right @deregtd, I did some tests and confirmed the stack trace information is not stored in the error itself so this change can be very helpful.
While reviewing the code I noticed the panic is not properly wrapped when it implements the error interface, so i created a PR with a fix for that which also includes the change to add the stack trace to the error message. Please take a look at it and let me know if that would work for you #96.
Since you came up with this idea, feel free to replicate the changes I made in your own branch so we can merge this PR instead of mine. That way you'll be the author of these changes 😉

[deregtd]

Haha I appreciate the offer, but I don't need the recognition. :). Just glad to help and eliminate my fork. :D I'll close this.