updater: Support max entry size in offline vuln load

[jvdm]

This PR allows clients to control the size of the offline importer entries. Up until now, the entry size was determined by the reference ID. Some updaters (eg. RHEL) fetch huge pages of vulnerabilities from their security source (eg. OVAL XML) into memory, and the JSON blob storage will give them the same reference ID, forcing clients to have enough memory to load them.

Other alternatives are possible, but I concluded that this one was less intrusive and had the lowest impact on existing clients.

Returning multiple entries with the same fingerprint should be OK as long the clients are the only ones performing the updates, and the state of the operations is retrieved before updating, which is what the offline importer does

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 55.82%. Comparing base (813e7cc) to head (5149ac2).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1304      +/-   ##
==========================================
+ Coverage   55.78%   55.82%   +0.03%     
==========================================
  Files         265      265              
  Lines       16547    16557      +10     
==========================================
+ Hits         9231     9243      +12     
+ Misses       6357     6356       -1     
+ Partials      959      958       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.