De ITN: decouple cardinal grammar from TN

[adelina23dunina]

What does this PR do ?

Rebuilds the German ITN cardinal grammar as a standalone WFST with its own TSV data, so it no longer depends on inverting the German TN cardinal tagger and verbalizer.

Details:

• CardinalFst in both the tagger and the verbalizer now take no arguments. The grammar is
  built from new tables in de/data/cardinal/ (zero, digits, irregular_teens, teens,
  tens) instead of from tn_cardinal_tagger / tn_cardinal_verbalizer.
• Adds de/graph_utils.py and de/utils.py to the ITN submodule.
• Output now carries German thousands separators, e.g. eintausend -> 1.000 rather than 1000.
• Numbers below thirteen are still left spelled out, as before.
• Keeps the attributes the other German semiotic classes read off the cardinal tagger, so
  date/decimal/fraction/measure/money/ordinal/telephone are unaffected:
  graph_no_exception, graph_hundred_component_at_least_one_none_zero_digit,
  optional_minus_graph.
• Bumps DE_TN_CACHE in the Jenkinsfile so CI rebuilds the grammars instead of reusing
  stale compiled FARs.

Before your PR is "Ready for review"

Pre checks:

• Have you signed your commits? Use git commit -s to sign.
• Do all unittests finish successfully before sending PR?
  1. pytest or (if your machine does not have GPU) pytest --cpu from the root folder (given you marked your test cases accordingly @pytest.mark.run_only_on('CPU')).
  2. Sparrowhawk tests bash tools/text_processing_deployment/export_grammars.sh --MODE=test ...
• If you are adding a new feature: Have you added test cases for both pytest and Sparrowhawk here. — updated tests/nemo_text_processing/de/data_inverse_text_normalization/test_cases_cardinal.txt, which feeds both test_cardinal.py and test_sparrowhawk_inverse_text_normalization.sh
• Have you added __init__.py for every folder and subfolder, including data folder which has .TSV files? — added de/data/__init__.py and de/data/cardinal/__init__.py
• Have you followed codeQL results and removed unused variables and imports (report is at the bottom of the PR in github review box) ?
• Have you added the correct license header Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved. to all newly added Python files?
• If you copied nemo_text_processing/text_normalization/en/graph_utils.py your header's second line should be Copyright 2015 and onwards Google, Inc.. See an example here.
• Remove import guards (try import: ... except: ...) if not already done. — none present in inverse_text_normalization/de/
• If you added a new language or a new feature please update the NeMo documentation (lives in different repo). — n/a, no new language and no user-facing API change
• Have you added your language support to tools/text_processing_deployment/pynini_export.py. — n/a, German is already registered there

PR Type:

• New Feature
• Bugfix
• Documentation
• Test

[mgrafu]

tsv file instead of hardcoding

[mgrafu]

let's use the closure instead of adding .ques

[mgrafu]

year