Update README with latest bm25 metrics and tokenization strategies

Update bm25 subset from local auto-best outputs

Add temporary construction notice below metadata

Normalize tokenization strategy label to wordseg

Add per-split selected BM25 tokenization strategy table

Consolidate bm25 note into Data structure and remove duplicate note in performance section

Unify README structure and add Data structure bm25 details

Normalize bm25 split names to language codes

Normalize qrels split names to language codes

Normalize queries split names to language codes

Normalize corpus split names to language codes

Add BM25 nDCG@10 to model results table and note bm25 subset

Add bm25 subset config from nano-ir-eval auto-best outputs

Update README.md

Align README split names with configs (NanoMIRACL_xx)

Fix configs metadata: use valid split names for datasets parser

Fix README front matter: add configs for datasets loader compatibility

Update README with links and ./nano_eval.py

Add NanoMIRACL README (new structure)