fix: make set_active_gateway concurrency safe

[maan2003]

fixes #4110

just put a lock around it for now

return error on concurrent calls

[maan2003]

i am conflicted with return Error and using lock

error might help you find concurrent calls with set_active_gateway, but lock feels more correct semantically

[elsirion]

Why not get rid of set_active_gateway altogether?

[elsirion]

i am conflicted with return Error and using lock

Why not auto-commit, if you want to avoid the error case? I don't think it makes sense though, just as the lock version.

In order of preference:

• Remove the active GW concept
• Fail on concurrent writes
• Autocommit
• Locking

[maan2003]

we still have to keep active gateway in fedimint-cli because of devimint

[justinmoon]

Relevant discussion in #3844

[m1sterc001guy]

Propagating the error up is IMO the best short term solution

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (df4bd8e) 58.01% compared to head (b4227a7) 58.05%.
Report is 30 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4162      +/-   ##
==========================================
+ Coverage   58.01%   58.05%   +0.03%     
==========================================
  Files         192      192              
  Lines       42990    42990              
==========================================
+ Hits        24941    24957      +16     
+ Misses      18049    18033      -16     

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