This was found locally by adding a small batch size to the WF helper code, which passed for most tests except for spill_to_disk.js which failed for the following reason:
[js_test:spill_to_disk] assert: command failed: {
[js_test:spill_to_disk] "ok" : 0,
[js_test:spill_to_disk] "errmsg" : "Executor error during getMore :: caused by :: Failed to write to disk because WiredTigerRecordStore::insertRecord 95: Operation not supported",
[js_test:spill_to_disk] "code" : 5643002,
[js_test:spill_to_disk] "codeName" : "Location5643002"
[js_test:spill_to_disk] }