Do not compare arrays with strings in make_initial_point_expression

[ricardoV94]

This gets rid of numpy FutureWarnings when we set the initval to an array:

import numpy as np
import pymc as pm

with pm.Model() as m:
  x = pm.Normal('x', size=10, initval=np.arange(10))
m.recompute_initial_point()

[codecov]

Codecov Report

Merging #5133 (510d7b8) into main (8950f21) will increase coverage by 9.71%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #5133      +/-   ##
==========================================
+ Coverage   67.88%   77.60%   +9.71%     
==========================================
  Files          88       88              
  Lines       14098    14100       +2     
==========================================
+ Hits         9571    10942    +1371     
+ Misses       4527     3158    -1369     

Impacted Files	Coverage Δ
pymc/initial_point.py	100.00% <100.00%> (ø)
pymc/step_methods/hmc/integration.py	78.84% <0.00%> (-3.85%)	⬇️
pymc/parallel_sampling.py	86.33% <0.00%> (-1.00%)	⬇️
pymc/tests/sampler_fixtures.py	90.51% <0.00%> (-0.73%)	⬇️
pymc/step_methods/hmc/nuts.py	95.00% <0.00%> (-0.63%)	⬇️
pymc/step_methods/hmc/base_hmc.py	90.32% <0.00%> (ø)
pymc/backends/base.py	86.25% <0.00%> (+0.76%)	⬆️
pymc/step_methods/hmc/hmc.py	92.30% <0.00%> (+1.92%)	⬆️
pymc/distributions/discrete.py	98.02% <0.00%> (+2.30%)	⬆️
pymc/util.py	74.35% <0.00%> (+2.56%)	⬆️
... and 22 more

[michaelosthege]

Test failures are unrelated (flaky tests?).

I remember that you (?) mentioned this problem somewhere, but it looks like there's no corresponding issue?

Anyway, your changes fix it so that's good.

[ricardoV94]

The failures are coming from #5131 because the Metropolis proposals are no longer following a global seeding, but several tests in test_step are based on this assumption.