I need 3 modifications of Anki 10k

[Expertium]

1. Vaibhav has an idea how to use real interval lengths for short-term scheduling. However, first I need to develop a formula that fully utilizes fractional interval lengths. So the first dataset that I need is the current Anki 10k, but with interval lengths recalculated from elapsed_seconds, such that same-day reviews have an interval length between 0 and 1. Relevant: https://forums.ankiweb.net/t/due-column-changing-days-from-whole-numbers-to-decimals-in-scheduling/52213/6?u=expertium, https://forums.ankiweb.net/t/due-column-changing-days-from-whole-numbers-to-decimals-in-scheduling/52213/12?u=expertium

2. I also want to incorporate sibling reviews into FSRS. Relevant: Discussing the new dataset and benchmark #129 (comment). I recommend making a new branch of the optimizer, since the optimizer will now need 3 values as inputs: interval length, grade and sibling_review.

3. Finally, if both ideas prove to be fruitful, I will need a dataset with both intervals lengths as floating point numbers and with the extra data from sibling reviews.

[L-M-Sherlock]

For the first modification, it has been implemented:

srs-benchmark/other.py

Lines 1980 to 1981 in f49dc94

	if SECS_IVL:
	df["delta_t"] = df["elapsed_seconds"] / 86400

I also want to incorporate sibling reviews into FSRS.

I haven't figured out how to implement it with minimum refactoring. Please wait for a few weeks.