[r] Improve error printing in call_peaks_macs

[bnprks]

I noticed that call_peaks_macs wasn't printing out helpful error messages because they get trapped inside the mclapply call sometimes. (This came up because I accidentally broke macs in my local virtualenv, but when tests failed there weren't good error messages)

Previous output when MACS hit an error while running peak calling:

Warning message:
In parallel::mclapply(file_names, function(shell_file) { :
  2 function calls resulted in an error

New output when MACS hits an error while running peak calling:

2024-12-20 16:01:21  Error running MACS for cluster: a_group
MACS log file written to: /tmp/RtmpwYXpma/file30a8322a9d7527/macs-test/output/a_group/log.txt
2024-12-20 16:01:21  Error running MACS for cluster: b_group
MACS log file written to: /tmp/RtmpwYXpma/file30a8322a9d7527/macs-test/output/b_group/log.txt
Error in `call_peaks_macs()`:
! MACS calls encountered 2 failures
• See error logs listed above
Run `rlang::last_trace()` to see where the error occurred.

The problem here is that if we throw an error inside a mclapply call, then it just reports there was some error but the text never gets printed

[immanuelazn]

This is a good catch. I recall a previous conversation in which we were tyring to fix this, and the log_progress() conditional for if (.Platform$GUI == "RStudio") ended up resolving it. It was probably an oversight on the first pass with not checking whether stop messages were also affected by multithreaded mclapply. It might make sense to flag message() and stop this to the r-source devs as these issues have existed for more than 10 years now.