chore(linear_algebra/dimension): more same-universe versions of `is_basis.mk_eq_dim` (#4591)
While all the `lift` magic is good for general theory, it is not that convenient for the case when everything is in `Type`.
* add `mk_eq_mk_of_basis'`: same-universe version of `mk_eq_mk_of_basis`;
* generalize `is_basis.mk_eq_dim''` to any index type in the same universe as `V`, not necessarily `s : set V`;
* reorder lemmas to optimize the total length of the proofs;
* drop one `finite_dimensional` assumption in `findim_of_infinite_dimensional`.