The original suspicion was that txn_test13 ran out of memory and crashed on arm64 small machines so we temporarily increased the size of the machine (WT-8788). Historically, the same test has been running fine on x86 machines with the same amount of memory, but recent failures on Intel have been seen during the work on WT-10017 and increasing the RAM size appears to fix the issue there too. 

There appear to be two failures in the patch build log file in the following tests (see details in comments below):

• test_txn13.test_txn13.test_large_values(integer-row.4gb) and
• test_sweep04.test_sweep04.test_big_run

both of which seem to be generated as a result of removing  'distros: ubuntu2004-arm64-large' from evergreen.yml and thereby switching to a smaller instance.

The smaller instance has less RAM (16Gb rather than 32Gb), less CPU cores (4 rather than 8), and less disk space (50Gb rather than 400Gb):

• It appears that the smaller RAM triggers the failure in test_txn13. 
• It appears that having less CPU cores reduces performance meaning that the test_sweep04.py test fails.