Ad-hoc handling of is_float/is_integer guards

Summary:
In the very specific case
```Erlang
foo(I) when is_integer(I) -> ...
foo(F) when is_float(F) -> ...
```
ignore negative type info issued from the former clause in occurrence typing, so as to avoid a falsely empty type in the latter.

The logic is based around comparing the current guard with the following ones, if it is of the form `is_integer` or `is_float` and the dual is present after, then the negative info is ignored.
This negative info will however be kept after the second guard, so as to still properly refine number types.

This needs some comparison of variables, so the code had to be refactored to keep the alias maps available for every clause.

Reviewed By: ilya-klyuchnikov

Differential Revision: D59518988

fbshipit-source-id: 7af4233d48ea0b8c28b2a4c79fc381cec9c6ed87

crates/elp/src/resources/test/eqwalizer_tests/options/uncovered_clauses.pretty

@@ -38,20 +38,4 @@ error: clause_not_covered
 
 See https://fb.me/eqwalizer_errors#clause_not_covered
 
-error: clause_not_covered
-   ┌─ options/src/uncovered_clauses.erl:37:1
-   │
-37 │ uncovered_number(F) when is_float(F) -> ok;
-   │ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Clause is not covered by spec
-
-See https://fb.me/eqwalizer_errors#clause_not_covered
-
-error: clause_not_covered
-   ┌─ options/src/uncovered_clauses.erl:42:1
-   │
-42 │ uncovered_number_2(I) when is_integer(I) -> ok;
-   │ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Clause is not covered by spec
-
-See https://fb.me/eqwalizer_errors#clause_not_covered
-
-7 ERRORS
+5 ERRORS