Give users an option for float 32 as default

[timcallow]

It may be the case that, for many calculations, float 32 arithmetic provides a reasonable accuracy. Of course, this also reduces the storage load and memory requirements during matrix diagonlizations.

Float 64 is kept as the default, but this PR introduces float 32 as an option.

[codecov]

Codecov Report

Merging #189 (6ecf2c6) into develop (8901717) will increase coverage by 0.14%.
The diff coverage is 100.00%.

@@             Coverage Diff             @@
##           develop     #189      +/-   ##
===========================================
+ Coverage    91.66%   91.80%   +0.14%     
===========================================
  Files           14       14              
  Lines         2364     2380      +16     
===========================================
+ Hits          2167     2185      +18     
+ Misses         197      195       -2     

Impacted Files	Coverage Δ
atoMEC/config.py	100.00% <100.00%> (ø)
atoMEC/convergence.py	100.00% <100.00%> (ø)
atoMEC/numerov.py	88.03% <100.00%> (+0.76%)	⬆️
atoMEC/postprocess/conductivity.py	97.71% <100.00%> (ø)
atoMEC/postprocess/localization.py	93.96% <100.00%> (ø)
atoMEC/staticKS.py	95.55% <100.00%> (ø)
atoMEC/xc.py	98.40% <100.00%> (ø)

... and 1 file with indirect coverage changes