Prioritise user created catch-all dataset factory pattern over `{default}` from runners

[ankatiyar]

Description

From slack conversation with @noklam https://linen-slack.kedro.org/t/16713675/i-am-trying-to-create-a-catch-all-patterns-using-the-dataset#65f8bddf-cb66-4b35-8f5d-c0850669b907

The catch-all factory pattern added by the user in their catalog does not always override the default dataset set by the runner - This works as expected ✅ -

"{catch_all}":
  type: pickle.PickleDataset
  filepath: data/06_models/{catch_all}.pickle

However, this pattern doesn't ❌ -

"{nok}":
  type: pickle.PickleDataset
  filepath: data/06_models/{nok}.pickle

Context

This behaviour was introduced in https://github.com/kedro-org/kedro/pull/3332 and was intentional and is documented, however, it is a little bit strange since for a user defined "catch-all" pattern to be used instead of the default dataset defined by the runner, which is {default} has to be ranked higher in terms of priority (alphabetically) to be picked up. That limits the names of the catch all patterns (should begin with a/b/c etc).

Expected Result

• Should always use the catch all pattern defined by the user (if there is one) instead of the default provided by the runner
• {default} can still be a protected catch-all pattern name that can only be used by the runner
• We should add a regression test for this behaviour. The catch-all patterns behave as expected in 0.18.14 but not with 0.19.x