refactor MetcalfScoring methods (#254)

* update abc.py to change `npl` to class attribute

It might lead to inconsistent `npl` if both `__init__` and `setup` take `npl` as parameter.

* remove `calc_score` method

This method is just a wrapper on `_calc_raw_score` and `_calc_mean_std`.
Better to move its code to `setup` method.

* update `setup` to use `cls.npl`

-  use `cls.npl`
- remove setup code from cache

* rename objtype to type

* change the format of raw score data frame

* Update `get_links` to use new format of dataframe

* Update metcalf_scoring.py

* change LINK_TYPES to enum LinkType

Update metcalf_scoring.py

* update `_get_links`

* add type variable ObjectType

* change `obj_type` from a string value to actual data type

* remove unused function

* clean `get_links`

* update docstrings for class attributes

* use hashable object itself as index or column of dataframe

E.g. use `GCF` instead of `GCF.id` as the index or column of a dataframe. This will skip the steps to look up actual object with a given id.

* update the code for calculating standardised score

* update unit tests

* Apply suggestions from code review

Co-authored-by: Giulia Crocioni <55382553+gcroci2@users.noreply.github.com>

---------

Co-authored-by: Giulia Crocioni <55382553+gcroci2@users.noreply.github.com>

src/nplinker/scoring/abc.py

@@ -21,14 +21,7 @@ class ScoringBase(ABC):
     """
 
     name: str = "ScoringBase"
-
-    def __init__(self, npl: NPLinker):
-        """Initialize the scoring method.
-
-        Args:
-            npl: The NPLinker object.
-        """
-        self.npl = npl
+    npl: NPLinker | None = None
 
     @classmethod
     @abstractmethod