Get the checksum file from archive.apache.org

[jcoyne]

Previously, when a mirror was set it was trying to get the checksum from
www.us.apache.org which only had the checksums for the three supported
versions.

[jrochkind]

We just ran into this and spent some time debugging it ourselves before realizing there was already a PR here with a fix.

@cbeer is there any chance of getting this merged and included in a bugfix release? Is there anyone but @cbeer who can do that on this project?

[jrochkind]

@jcoyne Does this actually address the same thing as #114 , that's been there since January?

[dkinzer]

I think there might be something else going on with the checksum code. We tried overriding the checksum path url and it still failed even though sha1s were the same. The only workaround for us seems to be to ignore checksum altogether: tulibraries/tul_cob#755

[jrochkind]

I'm confused about the difference between ignore_checksum: true that @dkinzer is using, and valid: false that we are to do the same thing. Maybe because @dkinzer is also passing a checksum: value (then telling solr_wrapper to ignore it?) and we aren't (so need to tell solr_wrapper not to try to fetch it?)?

Regardless, I'd really like if this problem was actually fixed in a solr_wrapper release, so we didn't all have to find confusing workarounds.

@cbeer is there any chance of a review/merge/release? Pretty please?