Stop assuming all upperclassmen are on every packet

This line makes the assumption that every single upperclassman is a required signature on every packet. This may not be the case, due to e.g. changes in membership state that don't influence earlier packets. This results in a confusing list of signature counts when an upperclassman is required on only some of the packets, since they are listed as `$num_signed / $num_total` in the active section. This aught to be `$num_signed / $num_can_sign_as_upper`
https://github.com/ComputerScienceHouse/packet/blob/f61cd2b5b616e3b7f35fad95b515313ea2d50bcb/packet/templates/upperclassmen_totals.html#L39

This will probably require injecting more data into the upperclassmen dict here so that it's accessible in the template.
https://github.com/ComputerScienceHouse/packet/blob/f61cd2b5b616e3b7f35fad95b515313ea2d50bcb/packet/routes/upperclassmen.py#L51-L62

Start with packet/templates/upperclassmen_totals.html at line 39 and packet/routes/upperclassmen.py at lines 51-62. Trace how the upperclassmen data reaches the active section, then ensure the displayed denominator counts only packets each upperclassman can sign. Done means the active counts use num_can_sign_as_upper rather than num_total when membership state excludes earlier packets.

Written by the indexing model from the issue text.