Cleaning out archived patches

[Konstantin Ryabitsev]

Hi, all:

One way to slim down the DB size would be to turn archived patches into 
a simple mapping of patchid/msgid to the archival URL. A request to an 
archived patch would then be a simple redirect to the archival site, 
e.g.:

https://patchwork.kernel.org/patch/11865643 
 -> https://lore.kernel.org/r/a90452e4fe7c6064026edc8377debd456f54137d.1603944740.git.pavana.sharma@digi.com

Similarly,

https://patchwork.kernel.org/project/netdevbpf/patch/a90452e4fe7c6064026edc8377debd456f54137d.1603944740.git.pavana.sharma@digi.com/
 -> https://lore.kernel.org/r/a90452e4fe7c6064026edc8377debd456f54137d.1603944740.git.pavana.sharma@digi.com

The table of archived patches would simply be:

patch_id | proj_name | msgid    | archive_url
11865643 | netdevbpf | a9045... | https://lore.kernel.org/...

None of the above values would be foreign keys, which would allow 
cleaning up old patches and old projects without breaking links.

This should help dramatically reduce the number of rows involved in most 
operations.

What do you think?

-K