Minor trace chemistry transfer should log row count after dedupe #637

Transfer logs show e.g. "Loaded 29376 cleaned" but "Upserting 1000 MinorTraceChemistry records". The only way to get 1000 rows is either (1) deduplication by nma_GlobalID leaving 1000 unique IDs, or (2) many rows skipped in _row_to_dict. There is no logging that distinguishes these.

Requested action: In transfers/minor_trace_chemistry_transfer.py, after building row_dicts and calling _dedupe_rows(rows), log both len(row_dicts) and len(rows) (e.g. "Converted N rows to dicts, M unique by nma_GlobalID"). This makes it clear whether the reduction is from dedupe or from skipped rows and avoids confusion about whether the transfer is truncating data.