analyze-reader-error.yaml intermittent failure

[mattnibs]

Observed failure here: https://github.com/brimdata/brimcap/runs/3127153671?check_suite_focus=true

From the logs:

 --- FAIL: TestBrimcap/cmd/brimcap/ztests/analyze-reader-error (1.12s)
26
            ztest.go:398: /Users/runner/work/brimcap/brimcap/cmd/brimcap/ztests/analyze-reader-error.yaml: stderr: regex \{"type":"warning","warning":"\.\/badoutput.sh: .*bad\.json: parse error: parsing string literal"\}
27
                \{"type":"status","ts":\{"sec":\d+,"ns":\d+\},"pcap_read_size":737694,"pcap_total_size":737694,"records_written":7\}
28
                 does not match {"type":"warning","warning":"./badoutput.sh: /private/var/folders/24/8k48jl6d249_n_qfxwsl6xvm0000gn/T/analyze-reader-error329052898/wd1/bad.json: parse error: parsing string literal"}
29
                {"type":"status","ts":{"sec":1626893233,"ns":511948000},"pcap_read_size":737694,"pcap_total_size":737694,"records_written":6}
30
                {"type":"status","ts":{"sec":1626893233,"ns":586789000},"pcap_read_size":737694,"pcap_total_size":737694,"records_written":7}

Looks like the analyze in this case took a little longer than usual and emitted two stats updates where the test only expects one. Let's fix the test to be more robust.