Improve on error handling when column is not found

[ivergara]

When a column specified in a constraint is not found, an unmanaged exception is shown that doesn't help in understanding what's the problem

______ test_constraint[UniquesSubset::X.X.X] ______

constraint = <dbcheck.constraints.uniques.UniquesSubset object at 0x7f9be45be730>
engine = Engine(postgresql://XXXX:***@XXXXXXXXXXXX:5432/health)

    @pytest.mark.parametrize("constraint", get_constraints(SCHEMA, OLD_SCHEMA), ids=idfn)
    def test_constraint(constraint, engine):
>       test_result = constraint.test(engine)

test_data.py:59: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
X/lib/python3.8/site-packages/dbcheck/constraints/base.py:178: in test
    value_factual = self.get_factual_value(engine)
X/lib/python3.8/site-packages/dbcheck/constraints/base.py:110: in get_factual_value
    factual_value, factual_selections = self.retrieve(engine, self.ref)
X/lib/python3.8/site-packages/dbcheck/constraints/uniques.py:94: in retrieve
    uniques, selection = db_access.get_uniques(engine, ref)
X/lib/python3.8/site-packages/dbcheck/db_access.py:377: in get_uniques
    selection = ref.get_selection(engine).alias()
X/lib/python3.8/site-packages/dbcheck/db_access.py:150: in get_selection
    [table.c[column_name] for column_name in self.columns]
X/lib/python3.8/site-packages/dbcheck/db_access.py:150: in <listcomp>
    [table.c[column_name] for column_name in self.columns]
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <sqlalchemy.sql.base.ImmutableColumnCollection object at 0x7f9be365e5e0>
key = 'l'

    def __getitem__(self, key):
        try:
>           return self._index[key]
E           KeyError: 'l'

KeyError: 'l' in this case is not very useful, l being the name of the column not found.