[polars] Sort joined frame by Date so yearly returns are correct

`read_data_polars` builds the combined price table with a chain of
`join(..., how='full', coalesce=True)` calls. Polars does not guarantee
join output row order: keys present on only one side are appended to the
end of the frame rather than slotted into place.

The Exercise 2 solution then takes `drop_nulls().first()/.last()` within
`group_by('year')`, which assumes rows are date-ordered. With the index
data that assumption breaks, because the Nikkei trades a different
holiday calendar from the US indices, so Nikkei-only dates land at the
end of the frame. The US columns are null on those appended rows and so
`drop_nulls()` skips them, which confines the damage to the Nikkei
column and let it pass review unnoticed.

Verified against Yahoo Finance data: the fix changes 51 of 204 yearly
return cells, all in the Nikkei column. Several were badly wrong and
some had the wrong sign (2017 read -6.42% against an actual +16.18%,
2020 read +3.79% against +18.27%). The code ran green throughout and
printed plausible numbers, which is the worst failure mode for a
published lecture.

Also in this commit:

- Add a note on join ordering, which reinforces the lecture's own
  "no index, no automatic alignment" theme and is a genuine gotcha
  when migrating from pandas.
- Drop an orphaned `[^mung]` footnote carried over from pandas.md. The
  word "munging" appears nowhere in this lecture, so the definition
  rendered as nothing at all.
- Cite the official Polars migration guide for the no-index discussion
  instead of a personal Medium post, for authority and longevity.
- Drop the unused `(pd-series)=` label from pandas.md. Nothing
  references it, so the PR now touches a single lecture.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: mmcky

lectures/pandas.md

@@ -78,7 +78,6 @@ You can think of a `Series` as a "column" of data, such as a collection of obser
 
 A `DataFrame` is a two-dimensional object for storing related columns of data.
 
-(pd-series)=
 ## Series
 
 ```{index} single: Pandas; Series

lectures/polars.md

@@ -86,7 +86,7 @@ s
 Unlike {doc}`pandas <pandas>` Series, Polars Series have no row index.
 Polars is column-centric --- data access is managed through column expressions
 and boolean masks rather than row labels.
-See [this blog post](https://medium.com/@luca.basanisi/understand-polars-lack-of-indexes-526ea75e413) for more detail.
+See the [Polars migration guide for pandas users](https://docs.pola.rs/user-guide/migration/pandas/) for more detail.
 ```
 
 Polars `Series` are built on top of [Apache Arrow](https://arrow.apache.org/) arrays and support many familiar operations
@@ -658,11 +658,20 @@ def read_data_polars(ticker_list,
         result = result.join(
             df, on='Date', how='full', coalesce=True
         )
-    return result
+    return result.sort('Date')
 
 ticker = read_data_polars(ticker_list)
 ```
 
+```{note}
+Polars joins do not guarantee the order of the output rows --- keys that match
+only one side are appended rather than slotted into place.
+This is the same "no index, no automatic alignment" theme from above: with no
+row labels to align on, ordering is something we ask for explicitly.
+Hence the `sort('Date')` before returning, which any later
+`first()`/`last()` calculation relies on.
+```
+
 Complete the program to plot the result as a bar graph.
 
 ```{exercise-end}
@@ -796,5 +805,3 @@ plt.show()
 
 ```{solution-end}
 ```
-
-[^mung]: Wikipedia defines munging as cleaning data from one raw form into a structured, purged one.