eum.py: ~120 lines of commented-out, half-translated EUM filter API stubs — port or delete

`mikecore/eum.py` carries roughly 120 lines of commented-out, half-translated C# declarations — from about line 1904 to about line 2025 — each carrying a `# TODO: Pending` marker. Nine of the module's TODOs live inside them. They look like this:

```python
# #/
# #/ TODO: Pending
# #/
# @staticmethod
# def eumGetFilterCount(); -> int:
#
# private static bool _eumGetFilterSeq(SeqNo: int, out FtKey: int, out IntPtr lpFtDesc);
```

They are not valid Python even uncommented — the signatures are a mix of Python and C# (`def eumGetFilterCount(); -> int:`, `private static bool`, `out IntPtr`), so they are a transcription of the original API surface rather than code that ever ran.

The affected group is the EUM filter/sequence family (`eumGetFilterCount`, `eumGetFilterSeq`, and neighbours).

Worth deciding once, either way:

- **Port them**, if anything needs to enumerate EUM filters from Python; or
- **Delete them**, and note in the module docstring that the filter API is not wrapped.

Either outcome removes nine TODOs and about 120 lines of noise from the largest module in the package. Leaving them is the only option that keeps implying work is planned when none is.

Untersuche den auskommentierten EUM-Filter-/Sequenzblock in mikecore/eum.py im Bereich der Zeilen 1904–2025 sowie die umgebende gewrappte API, um festzustellen, ob Python-Aufrufer eine Filteraufzählung benötigen. Erledigt ist die Aufgabe, wenn entweder die Deklarationen durch gültige Python-Wrapper ersetzt werden oder der Block und seine neun TODOs entfernt und die nicht gewrappte Filter-API im Modul-Docstring dokumentiert wird.

Vom Indexierungsmodell aus dem Issue-Text verfasst.