Laziness of boolean operators about biscuit

[Tipnos]

From my point of view this PR is bound with PR #130 (Heterogeneous == ). I think if you decide to not enforce that "strict evaluation, where having a non-computable expression aborts the whole evaluation". I think it's important also to accept "equality on heterogeneous terms" to stay consistent. The specification will be more intuitive.

Moreover the absence of type checking and compilation is an argument in favor of that. It would be weird to have unexpected "type checking" errors at runtime and no tools to prevent them upstream.

[Geal]

I agree, it should be lazy, that's the most commonly expected behaviour

[divarvel]

i have a proof of concept here biscuit-auth/biscuit-rust#188

the proof-of-concept makes them non-strict rather than lazy. The difference is subtle (and not observable from the outside in the absence of side-effects), but still important to keep in mind.