Middleware

[rtrox]

Description of the change

Adds middleware for all collectors. Two middlewares are added:

1. Metrics middleware -- collects metrics related to the scrape itself (specifically latency & request count).
2. Recovery middleware -- after panic: runtime error: index out of range #138, I added some recovery specifically in the client, which I think is still valuable, so that we can log json blobs which fail to unmarshal. But for the sake of reliability, we should really add a catch-all recover middleware. With this middleware, rather than panics cause a crash, they'll instead log, and return a 500 to prometheus.

Since I had to add a writer wrapper to collect the status code for requests, I went ahead and added it to the logger middleware we already had, too, as a QoL change.

Benefits

1. Prevents panic-crashes
2. Exports metrics about scrapes.

Possible drawbacks

Applicable issues

• fixes #

Additional information

[rtrox]

This is ready to go, but draft until #141 lands, so this is easier to review.