Safer keys for memcached

[lferran]

What this PR does

Hashes cache keys before sending it to memcached, to avoid having keys that aren't valid.

Pros:

• right now, keys include db_id, and possibly the uuid of the object + container_id. The length of the key is not bounded right now. This PR fixes it in a simple way.

Cons:

• Adds a little bit of extra cpu
• Inspecting cache keys will be more difficult
• Hashing the whole key could introduce cross-container collisions, but the probability is very low. Assuming a cache of 10Gi:

prob_key_collision = 1.0 / 15**56
n_cache_keys = 10 * 2**30

prob_key_collision * n_cache_keys = ~ 1.5e-56

[codecov-io]

Codecov Report

Merging #1019 into 5.x will increase coverage by 0.1%. The diff coverage is 100.0%.

@@           Coverage Diff           @@
##             5.x   #1019     +/-   ##
=======================================
+ Coverage   94.7%   94.7%   +0.1%     
=======================================
  Files        296     296             
  Lines      27198   27219     +21     
=======================================
+ Hits       25750   25771     +21     
  Misses      1448    1448             

Impacted Files	Coverage Δ
guillotina/contrib/memcached/driver.py	85.0% <100.0%> (+0.4%)	:arrow_up:
...uillotina/tests/memcached/test_memcached_driver.py	100.0% <100.0%> (ø)

[vangheem]

5.3.58 released