Problem with conditional prior dict and log probabilities?

[KonstantinLeyde]

If I define the following conditional prior dict, the log_prob raises an error:

from bilby.core.prior import ConditionalPriorDict, PowerLaw, ConditionalPowerLaw, Constraint
from bilby.gw.conversion import generate_mass_parameters

def secondary_condition_function(reference_params, mass_1_source):
    return dict(minimum=reference_params['minimum'], maximum=mass_1_source)

p = dict(
    alpha=1,
    beta=2,
    minimum=3,
    maximum=10,
)

prior = ConditionalPriorDict(
    {'mass_1_source': PowerLaw(
            alpha=-p["alpha"],
            minimum=p["minimum"],
            maximum=p["maximum"],
        ),
    "mass_2_source": ConditionalPowerLaw(
            secondary_condition_function,
            alpha=p["beta"],
            minimum=p["minimum"],
            maximum=p["maximum"],
    ),
    "mass_ratio": Constraint(minimum=0.125, maximum=1.0),
    },
conversion_function=lambda x: generate_mass_parameters(x, source=True),
)

s = prior.sample(size=5)


# gives an error
log_prob = prior.ln_prob(s)
print(log_prob.shape)

# # works as expected
# log_prob = prior.ln_prob(s, axis=0)
# print(log_prob.shape)

The error message I get is

    log_prob = prior.ln_prob(s)
  File "/opt/homebrew/Caskroom/mambaforge/base/envs/dingo-pop/lib/python3.10/site-packages/bilby/core/prior/dict.py", line 809, in ln_prob
    return self.check_ln_prob(sample, ln_prob)
  File "/opt/homebrew/Caskroom/mambaforge/base/envs/dingo-pop/lib/python3.10/site-packages/bilby/core/prior/dict.py", line 565, in check_ln_prob
    if self.evaluate_constraints(sample):
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

If I do not enforce the mass ratio constraint (and also delete the conversion_function), this works as expected (calculating the probability over all samples). Also, if I add axis=0, this returns the log probability for each sample, also not throwing an error. Is this behavior expected? Thanks!