[abhiishek340]

Closes #76.

Empty cells already parse correctly when the input's line breaks match lineBreak:

new csv.Parser { input = "A,B,C\r\n1,,3"; includeHeader = true }.parsed
// List(new { A = "1"; B = null; C = "3" })

The failure in #76 comes from line breaks. lineBreak defaults to "\r\n" (RFC 4180), and the input in the issue is a Pkl multi-line string, which uses "\n". With no "\r\n" in the input, it's parsed as a single row. The empty cell from the second line then lands in the header, and the as List<String>? cast fails with:

Expected value of type `String`, but got `null`.

This PR replaces that with a message that names the problem and, when the input contains "\n" but no lineBreak, how to fix it:

Column 4 of the header row is empty, so it can't be used as a column name. The input contains no "\r\n" line breaks, so it was parsed as a single row. If it uses "\n" line breaks, set `lineBreak = "\n"`.

A header with a genuinely empty column ("A,,C\r\n1,2,3") gets just the first sentence.

I didn't change the default line break or start accepting both, since that changes parsing semantics. Happy to do that instead if you'd prefer.

Changes:

• csv.pkl: check the header row for empty columns before the cast, plus a _lineBreakHint for the single-row case.
• tests/csv_test.pkl: new "Parsing with an empty header column" fact covering both messages and the lineBreak = "\n" case.
• PklProject: bump pkl.csv to 1.1.2. No other package depends on pkl.csv.

Testing: pkl test packages/pkl.csv/tests/csv_test.pkl passes with Pkl 0.32.1 (the version in libs.versions.toml) and with pkl main: 13 tests, 55 assertions. pkl format (0.32.1) reports no changes in the new code. The three lines it does flag in csv.pkl were already flagged on main, so I left them alone.

[bioball]

LGTM, thank you!