limit.overrideKey is computed incorrectly

[aarongable]

When parsing limit overrides, we first create a pointer to a limit{} struct:

boulder/ratelimits/limit.go

Line 159 in 1b7b9a7

limit := &v.limit

then, for each override ID, we set that limit's overrideKey field:

boulder/ratelimits/limit.go

Line 177 in 1b7b9a7

limit.overrideKey = joinWithColon(name.EnumString(), id)

then save the pointer to this limit into a map:

boulder/ratelimits/limit.go

Line 185 in 1b7b9a7

parsed[joinWithColon(name.EnumString(), id)] = limit

If a limit has 5 override IDs, the "overrideKey" field of that singular limit struct will get overwritten 4 times, resulting in only the last value actually being preserved in the limit struct. This will throw off metrics about override utilitzation:

boulder/ratelimits/limiter.go

Line 293 in 1b7b9a7

l.overrideUsageGauge.WithLabelValues(txn.limit.name.String(), txn.limit.overrideKey).Set(utilization)

[beautifulentropy]

In an offline discussion we decided to remove this an other metrics that we we're not finding useful.