[MRG] MAINT make hash deterministic

[glemaitre]

Reference Issue

closes #344
closes: #365

What does this implement/fix? Explain your changes.

Remove the fixed random_seed to compute the hash. Instead we pick-up equally spaced samples.
Since this is related, random_state has been deprecated whenever possible.

TODO:

• tests

Any other comments?

[codecov]

Codecov Report

Merging #374 into master will decrease coverage by 0.02%.
The diff coverage is 98.6%.

@@ Coverage Diff @@ ## master #374 +/- ## ========================================== - Coverage 98.74% 98.71% -0.03%  ========================================== Files 68 68 Lines 3899 3969 +70 ========================================== + Hits 3850 3918 +68  - Misses 49 51 +2

Impacted Files	Coverage Δ
imblearn/over_sampling/base.py	100% <ø> (ø)	⬆️
..._sampling/prototype_selection/tests/test_allknn.py	100% <100%> (ø)	⬆️
imblearn/ensemble/balance_cascade.py	100% <100%> (ø)	⬆️
imblearn/over_sampling/random_over_sampler.py	100% <100%> (ø)	⬆️
...prototype_selection/instance_hardness_threshold.py	100% <100%> (ø)	⬆️
...ampling/prototype_selection/tests/test_nearmiss.py	100% <100%> (ø)	⬆️
...prototype_selection/neighbourhood_cleaning_rule.py	100% <100%> (ø)	⬆️
imblearn/utils/validation.py	100% <100%> (ø)	⬆️
...n/tests/test_repeated_edited_nearest_neighbours.py	100% <100%> (ø)	⬆️
...g/prototype_selection/edited_nearest_neighbours.py	100% <100%> (ø)	⬆️
... and 18 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e00e7f8...ea95b50. Read the comment docs.

[glemaitre]

ping @chkoar if you can check this PR. I deprecated in the same time whenever the random_state was not useful.

[glemaitre]

Actually this a blocker to be able to develop what ever new :)

[chkoar]

The changes were trivial and it doesn't matter anymore cause it is merge but I would probably replaced that line

if 'random_state' in sampler_attr: samplers = [Sampler(random_state=0)] else: samplers = [Sampler()]

with this function call :D

[glemaitre]

I was not aware that it was that smart. Can you make a ‎PR to correct this, I think this is worth for readibility

[chkoar]

@glemaitre ok