Prometheus cleanup

[benjimin]

Some cleanup related to (or unearthed in pursuit of) #129

• Remove stale code relating to custom metrics (that separate different queries). This is a role for logs not metrics; Prometheus docs explicitly caution against unbounded label sets.
• Ensure gunicorn & prometheus configured not to leak memory, that is, ensure the shared memory filesystem used for communicating metrics between gunicorn worker processes is being cleaned up as those processes respawn (in concordance with documentation for prometheus flask exporter).

[codecov[bot]]

Codecov Report

Merging #133 (bc38b41) into master (b2c26f7) will increase coverage by 0.10%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #133      +/-   ##
==========================================
+ Coverage   66.15%   66.25%   +0.10%     
==========================================
  Files           8        8              
  Lines         647      643       -4     
==========================================
- Hits          428      426       -2     
+ Misses        219      217       -2     

Impacted Files	Coverage Δ
datacube_wps/startup_utils.py	77.27% <ø> (+4.19%)	:arrow_up:
datacube_wps/impl.py	73.07% <100.00%> (ø)