feat: reduce forecast horizon of questions to 1 year

[nikbpetrov]

Closes #296

• src/orchestration/func_resolve/main.py does the _io.download_and_read_question_set_file a few times now but should not be a problem, current commit matches existing patterns
• leaderboard: text update only; whether longer horizon Qs still get shown (first one to resolve in July '27) is a decision up for grabs!
• tests are assumed to be crappy (AI generated); can throw those away if you prefer; work on those will be in the dedicated testsuite work

[houtanb]

whether longer horizon Qs still get shown (first one to resolve in July '27) is a decision up for grabs!

Still show these. This is only a change for question sets going forward. Past question sets will continue to resolve their long horizon forecasts

Don't worry about tests now given this is all going to be redone

[nikbpetrov]

Tests pass.

[houtanb]

@nikbpetrov this is great!

Is there a reason you didn't want to update the _questions.jsonl file for acled and dbnomics? We can do that after the survey starts perhaps.

And, indeed, going to another conversation we had, perhaps the resolution dates don't belong on the questions file at all as they're not observations on the question. Instead we can add a field called "update frequency" (to handle series updated less than daily) and then set the resolution dates directly on question curation.

If we merge this as is, I'll create a new ticket for the rest of the cleanup for this after the survey.

[nikbpetrov]

oversight - other sources rebuild so i assumed that

anyhow, i like the resolution dates idea so if we are going that route, i'd skip over the immediate fix; if immediate fix for this round is needed, i'd probably opt for the dangerous life anyway: one-time script to update the 2 existsing _questions files