Replaces thread storage with request_store

[leonelgalan]

Fixes issue #197. Under load it was possible for a thread to handle a second request without clearing the variables in the thread. This caused log lines to have the incorrect values for : lograge_location.

@abrisse suggested we used request_store, after a brief examination of this gem it seems that it’s well maintained and it solves our problem.

[abrisse]

LGTM

[pxlpnk]

Looks good, I am just wondering about the sidekiq issue mentioned here:
https://github.com/steveklabnik/request_store/pull/54/files

I have no numbers though how many people use lograge with sidkiq.
@abrisse @leonelgalan do you run a sidekiq setup somewhere that utilises lograge?

[benlovell]

@pxlpnk I'm not sure that's a problem here. As far as I can tell Active Job and Sidekiq workers don't use our subscriber.

[pxlpnk]

Then this approach is definitely more sane and safe 👍

[benlovell]

Thanks! I'll cut a release today.

[pxlpnk]

I can't see a specific reason, why we would pin it to 1.0.

Lets update it to the latest: #219

[benlovell]

It's not pinned and will resolve to 1.3.2 as evidenced here:

https://github.com/guidance-guarantee-programme/pension_guidance/blob/76b5d68e3678357e7b2680b3b6dedf86b25d6970/Gemfile.lock#L303

[leonelgalan]

Sorry for the late response, I left it at '~> 1.0', because (1) it resolves to the latest, and (2) we don't use anything introduced after 1.0