Test figures badly wrong

[openstrike]

Describe the bug
At least for some dists, the numbers in the "Testers" field in the left nav are very different from those reported on matrix.cpantesters.org. This may be the same issue as #1870 or it may be something else. Merge these issues if desired.

To Reproduce
Steps to reproduce the behavior:

1. Go to https://metacpan.org/pod/Date::Christmas
2. See the Testers figures in the left column are 16 / 0 / 0
3. Click on the "Testers" link.
4. See the figures on http://matrix.cpantesters.org/?dist=Date-Christmas+1.02 are 847 / 0 / 2

Expected behavior
The total of pass / fail / unknown should match at least within +/- 10

Screenshots

Desktop (please complete the following information):

• OS: Linux 64-bit (Slackware 15)
• Browser Pale Moon
• Version 32.5.1 (64-bit)

Additional context
I believe the results on matrix.cpantesters.org. Somehow metacpan isn't picking these up correctly.

[openstrike]

The discrepancy seems most pronounced for those dists released longest ago. I haven't yet found any dist with a newest release over 15 years ago where the figures are even close. HTH.