@Sarah Moore great timing on this question — warehouse deployments in dense metal environments are genuinely one of the trickier scenarios to predict without hands-on testing, so I'll share what I know and be straight with you about where the uncertainty is.
The 50km figure is drone-to-drone with clean line of sight and no ground reflections, so yeah, you're right to mentally discount that heavily for your use case. In a dense metal warehouse with steel racking and a metal roof, you're dealing with multipath reflections, RF absorption from the structure itself, and likely some interference from motor drives and other industrial equipment on the 2.4GHz band. Realistically I'd expect usable range to drop significantly — we're probably talking tens of meters between nodes in the worst sections rather than hundreds. The receiver sensitivity on this module is genuinely good (that's where a lot of the outdoor range comes from), so it'll pull signal out of noise better than a commodity module would, but physics is physics when you've got steel everywhere.
On the data rate question — BLE at 2Mbps is a PHY layer setting, not something that automatically steps down the way WiFi does with MCS rate adaptation. So if the link degrades, you'll see it as packet errors and connection drops rather than a graceful rate reduction. If you need adaptive behavior in that environment, you'd handle it in firmware — either by falling back to 1Mbps or Coded PHY (which gives you much better range at lower throughput) depending on what the underlying chip supports. Worth digging into the datasheet specifically on that point.
For 200,000 sq ft of dense metal racking I'd honestly plan your node layout assuming you'll need infrastructure at reasonable intervals and think of the module's sensitivity as your margin of safety, not your reason to stretch the spacing. Worth doing a small pilot deployment in the actual space before you commit to a full node map — that'll tell you more than any spec sheet.
