Pedagogy: merge 00_introduction/50_pandas + 20_probability/30_pandas → 06_python_advanced/90_pandas

## Goal

Pandas is currently taught in two notebooks:

- `00_introduction/50_pandas.ipynb` (~40 cells): Series, DataFrame, vectorization, `.shift`, `.astype`, toy data
- `20_probability/30_pandas.ipynb` (~180 cells): `read_html`, `dropna`, `set_index`, `.loc`/`.at`, `.isin`, `.str`, seaborn, real Naver Finance data

The two are sequential (basics → application), not duplicative. But they live in mismatched folders — `20_probability/30_pandas` has no probability content — and the split into two files creates a discoverability gap.

## Scope

Create `06_python_advanced/90_pandas.ipynb` by merging the two:

1. **Series and DataFrame fundamentals** (from old 00/50)
2. **I/O and cleaning** — `read_csv` / `read_html`, `dropna`, `set_index`, `dtypes` (from old 20/30)
3. **Selection and querying** — `.loc`, `.at`, `.isin`, `.str` methods (from old 20/30)
4. **Visualization** — quick seaborn intro (from old 20/30)
5. **Vectorization vs loops** — concluding hook (from old 00/50)

Trim duplicated Series/DataFrame intro cells so each concept appears once.

Retire `00_introduction/50_pandas.ipynb` and `20_probability/30_pandas.ipynb` after the merge lands.

## Why

- One canonical pandas notebook, in the right folder.
- `20_probability/` stops claiming to teach pandas (it never did).
- Aligns with the broader `00_introduction` restructure (libraries move to `06_`).

## Acceptance

- [ ] `06_python_advanced/90_pandas.ipynb` exists and runs end-to-end.
- [ ] No content lost from either source notebook (each concept appears at least once).
- [ ] Both old notebooks removed (or stub-redirected for one CI cycle, then removed).