handleSummary called before all metrics are crunched

[na--]

A check I added on Friday (#1885) for xk6 JS and output extensions seems to have caught a problem with the end-of-test summary 😕 From https://github.com/loadimpact/k6/runs/2041232632:

2021-03-05T16:29:41.7585634Z + ./k6ext run --out outputtest=output-results.txt xk6-test.js
2021-03-05T16:29:41.7809953Z 
2021-03-05T16:29:41.7814813Z           /\      |‾‾| /‾‾/   /‾‾/   
2021-03-05T16:29:41.7815511Z      /\  /  \     |  |/  /   /  /    
2021-03-05T16:29:41.7816200Z     /  \/    \    |     (   /   ‾‾\  
2021-03-05T16:29:41.7816746Z    /          \   |  |\  \ |  (‾)  | 
2021-03-05T16:29:41.7817112Z   / __________ \  |__| \__\ \_____/ .io
2021-03-05T16:29:41.7817365Z 
2021-03-05T16:29:42.4474038Z   execution: local
2021-03-05T16:29:42.4475936Z      script: xk6-test.js
2021-03-05T16:29:42.4477250Z      output: test output extension
2021-03-05T16:29:42.4477895Z 
2021-03-05T16:29:42.4478714Z   scenarios: (100.00%) 1 scenario, 1 max VUs, 10m30s max duration (incl. graceful stop):
2021-03-05T16:29:42.4479781Z            * default: 5 iterations shared among 1 VUs (maxDuration: 10m0s, gracefulStop: 30s)
2021-03-05T16:29:42.4480493Z 
2021-03-05T16:29:42.4589016Z 
2021-03-05T16:29:42.4590100Z running (00m00.0s), 0/1 VUs, 5 complete and 0 interrupted iterations
2021-03-05T16:29:42.4591017Z default   [ 100% ] 1 VUs  00m00.0s/10m0s  5/5 shared iters
2021-03-05T16:29:42.6040590Z ++ cat output-results.txt
2021-03-05T16:29:42.6051809Z + OUTPUT_RESULT=10
2021-03-05T16:29:42.6052563Z + [[ 10 != \1\0 ]]
2021-03-05T16:29:42.6056159Z ++ cat summary-results.txt
2021-03-05T16:29:42.6064677Z + SUMMARY_RESULT=6
2021-03-05T16:29:42.6065207Z summary result was not as expected: 6
2021-03-05T16:29:42.6066851Z + [[ 6 != \1\0 ]]
2021-03-05T16:29:42.6067948Z + echo 'summary result was not as expected: 6'
2021-03-05T16:29:42.6068629Z + exit 12
2021-03-05T16:29:42.6096475Z ##[error]Process completed with exit code 12.

The test script emits the __ITER value for 5 iterations. It seems like the output received and processed all metrics (0+1+2+3+4==10), but the data handleSummary() got was missing its last value... 😕

I think I haven't broken this with #1869, but we should investigate, since it likely means we have another data race on our hands... 😞