Add Population Threshold RDD Italy dataset files - #18

**The following checklist is mandatory**

Your PR will be closed if you remove the checklist. Use of LLMs is
**strictly forbidden** for any part of this checklist (including for
improving language), and will result in a **ban** if we find any use of
LLMs.

### Your checklist for this pull request

- [x] Have you followed all the steps from our [Contributing
Guide](https://github.com/pgmpy/pgmpy/blob/dev/CONTRIBUTING.md)?
- [x] Does the PR fully address the linked issue and is within its
defined scope? If you are still working on the PR, mark it as draft.
- [x] Are all the GitHub Actions checks passing? If not, mark your PR as
draft while you fix it.

If you have used AI/LLMs for any assistance, please answer the following
questions. Please refer
[pgmpy#2622](pgmpy#2622) for an example of the
level of detail we expect:

- [x] Have you reviewed our [AI usage
policy](https://github.com/pgmpy/pgmpy/blob/dev/CONTRIBUTING.md#ai-usage-policy)?
- [x] Are you able to fully explain your changes? We expect you to fully
understand the algorithm and take full responsibility for any changes in
this PR.

- Please list the AI tool(s) used, along with the model and its version
used.

Grok 4.6 (xAI), used from the Grok Build TUI.

- Please describe in **detail** how and what you used AI assistance for?
Please outline your whole workflow with the AI tool.

I used Grok to search pgmpy#2620, related issues, and open PRs so I would not
take a dataset that already had an active PR. After I picked the Italy
table from the Eggers population-threshold RDD archive, I used it to
draft the BaseDataset subclass from
devtools/extension_templates/_dataset.py and the existing AngristKrueger
loader. I also used it to convert pop_devs_three_countries_v5.RData into
the tab-separated file, write the tag test, and add the two bibliography
entries. I read the resulting class, tags, and test myself before
opening this PR. I did not have it invent extra features or extra
datasets.

- What steps have you taken to verify that the changes correctly address
the issue? What edge cases have you considered? Other than running
tests, what else have you verified?

The Dataverse zip MD5 matched df3c2bd36a9f5fcd9d1e86d5b8e31f66. The
packaged Italy file is 10409 rows by 19 columns. I loaded that file
through PopulationThresholdRDDItaly.load_dataframe() locally (same
parser as the hub path) and checked n_samples / n_variables, missing
values, and macro_area as a category. I ran pytest -v
pgmpy/tests/test_datasets/test_datasets.py (10 passed) and pre-commit
ruff/format on the Python files. I did not add a live hub download test,
because the file is not on the Hub until the companion example_datasets
PR is merged. I left France and Germany out of this PR on purpose.

- Have you used AI for generating tests? Can you compress them into a
smaller number of tests without losing coverage?

Yes, for one tag/registry test. It only checks discovery, tag values,
the categorical column, and the missing-value marker. I did not add a
large generated suite.

### Issue number(s) that this pull request fixes
- Fixes pgmpy#2788
- Parent: pgmpy#2620 (this PR adds only the Population Threshold RDD Italy
table)

### List of changes to the codebase in this pull request
- Added PopulationThresholdRDDItaly in
pgmpy/datasets/population_threshold_rdd.py using the dataset extension
template (on-demand hub download + cache, metadata tags,
source/license/checksum in the docstring).
- Registered population_threshold_rdd_italy in ALL_DATASETS and added a
tag test.
- Added bibliography entries for the AJPS paper and the Harvard
Dataverse archive.

Hub data files (CC0, checksums in the README):
pgmpy/example_datasets#18