add setting PROMETHEUS_TOTAL_BYTES_BUCKETS for metric body size

[FirePanda169]

Hi
At work, I encountered a problem where our service has a release limitation due to the large number of buckets in some metrics. This limitation is due to the fact that a large number of buckets puts a heavy load on Prometheus.

More specifically, django_http_requests_body_total_bytes and django_http_responses_body_total_bytes
This is despite the fact that starting with le=2^17 in django_http_requests_body_total_bytes_bucket and le=2^24 in django_http_responses_body_total_bytes_bucket, there is no difference.
In fact, the difference between le=2^24 and le=2^17 in django_http_requests_body_total_bytes_bucket is 16 for 320k requests.

I propose enabling custom bucket sizes.