Switch `CompletionContext` to `dataclass`, use `cached_property`

Data class has narrower API than named tuple (does not expose item getters)
which means it is less likely that the downstream code would be broken in
the future if we introduce new attributes. Also, compared to other places
where memory-efficient named tuple is used, `CompletionContext` is only
created once per completion and thus does not require as low overhead.
For the same reason we can forgo slightly more memory-efficient
`@property @cache` stack and just use `@cached_property`.

IPython/core/completer.py

@@ -191,7 +191,8 b' import unicodedata'
 import uuid
 import warnings
 from contextlib import contextmanager
-from functools import lru_cache, partial
+from dataclasses import dataclass
+from functools import cached_property, partial
 from importlib import import_module
 from types import SimpleNamespace
 from typing import (
@@ -597,7 +598,8 b' class _JediMatcherResult(_MatcherResultBase):'
     completions: Iterable[_JediCompletionLike]
 
 
-class CompletionContext(NamedTuple):
+@dataclass
+class CompletionContext:
     """Completion context provided as an argument to matchers in the Matcher API v2."""
 
     # rationale: many legacy matchers relied on completer state (`self.text_until_cursor`)
@@ -626,13 +628,11 b' class CompletionContext(NamedTuple):'
     # If not given, return all possible completions.
     limit: Optional[int]
 
-    @property
-    @lru_cache(maxsize=None)  # TODO change to @cache after dropping Python 3.7
+    @cached_property
     def text_until_cursor(self) -> str:
         return self.line_with_cursor[: self.cursor_position]
 
-    @property
-    @lru_cache(maxsize=None)  # TODO change to @cache after dropping Python 3.7
+    @cached_property
     def line_with_cursor(self) -> str:
         return self.full_text.split("\n")[self.cursor_line]