Fix stale CSV header missing 'open' column in statistics_history.csv

[Chessing234]

The open field was added to the row schema in commit 06fdb21 (scripts/plot_statistics_history.py FIELDNAMES) but the CSV header on line 1 of data/statistics_history.csv was never updated, since the header is only written by update_history() when the file doesn't yet exist. Every row appended since 2025-12-12 has 11 values while the header names only 10 columns.

Verified with csv.DictReader (what the repo's own scripts use to read this file): the 11th value on each affected row lands under the None restkey instead of open, so row.get("open", ...) in generate_charts() never finds the recorded value and silently falls back to a computed default. This happens to match numerically right now (default = total_problems - total_solved), which is why it wasn't noticed, but the CSV is malformed for any other consumer (pandas, spreadsheets, etc.) and the real recorded 'open' values are being discarded on every read.

Fix: add open to the header row so it matches the 11 columns actually written. Confirmed with DictReader before/after that the 'open' key now parses correctly for all post-06fdb21 rows, and that older rows (pre-'open' schema) correctly get None for the missing field. Ran python scripts/validate.py — passes.

[Chessing234]

rebased onto main, and added the script side — adding open to the header alone breaks chart generation.

csv.DictReader creates a key for every column the header declares, so the existing row.get("open", ...) fallback never fires; the 350 rows written before that column get open: None and int(None) raises. running plot_statistics_history.py on this branch before the second commit: TypeError at line 134 in generate_charts, exit 1. after: exit 0.