Properly reset sequence if switching with multiple schemas

[ryanswood]

The recent change to fix table sequences does not account for multi-schema switching:

def default_sequence_name(table, _column)
  res = super
  # Code expects a single schema not an array
  schema_prefix = "#{Apartment::Tenant.current}." # => "[\"tenant1\", \"tenant2\"]."

thus sequences are not scoped correctly.

Before (incorrect):

> Apartment::Tenant.switch(['tenant1', 'tenant2']) { Foo.sequence_name }
=> tenant1.foos_id_seq

After:

> Apartment::Tenant.switch(['tenant1', 'tenant2']) { Foo.sequence_name }
=> foos_id_seq

[ryanswood]

Without resetting the sequence, the spec would pass if ran with the higher level spec with the same name because the sequence_name is cached on the model. I applied the same process to the other spec to avoid false positives depending on how RSpec orders the tests.

Let me know if there is a better way to do this.

[mnovelo]

@ryanswood would you mind rebasing?

[ryanswood]

@mnovelo Sorry for the delay. I wasn't feeling well for a bit. PR has been rebased!