Pi 5 stopping after 3 minutes and SDR in the wrong order

I don’t recall this numbering (device 0 SN 1004 etc.) and was surprised recently when I saw it.
There are a number of other reports in this forum (eg this recent one) that device 0 is SN 1000 which is what I expected and maybe hallucinated that I recalled seeing in prior runs of rtl_test (and kraken_test).

Did something change regarding this not-mattering serial enumeration or has it always been this way?