Disabling STP on a port basis seems possible, but I'd really reconsider the reasoning as to why you'd want to do that. You can disable it in the interface context (at least on the M4300's) using the 'no spannin-tree port mode' command in the CLI.
If a loop is caused by such a port it cannot be stopped to quickly become a broadcast storm tearing down larger segments of your network. It's like using a safety belt for the driver but not your passengers.
There are other ways if you run into issues because STP blocks a port for a short time before it becomes online like 'spanning-tree edgeport' and using RSTP or MSTP instead of the legacy STP protocol.
In your case and since you explained your topology later as in the initial post one 1 of the 2 links will be blocked by both STP and RSTP.
Technicaly you could consider MSTP - ignoring the originally vendor-specific PVST and RPVST/PVST+ modes - but then again, @schumaku 's suggestion of creating an LAG on both ends and tag all VLANs on the LAG interface is proven to work (for me) and provides proper redundancy where both links can be used by all VLANs.
However the Dell switch needs to support link aggregation with LACP. You might also have to fiddle out what mode (dynamic or static LAG) works with both vendors.
Keep your STP configuration simple and it will work reliably when loops occur. Don't try to outsmart it 😉
