Multithread versus single thread performance

I'm sorry but I haven't a clue where I should post this but help seems the most appropriate!

I'm afraid it is a multiple question with one major one at the centre so to speak.

Main question - should multi thread and single thread tests give similar download speeds?

Related questions -

1) My understanding of multi versus single thread is questionable at best. Is single thread a random file downloaded from a single server, so if that server is overloaded, that could slow down the 'flow' from that server all the way to final destination (ie my computer) thus giving a false impression of the ISPs performance?

2) Does multithread overcome this by literally throwing several random files from random servers across 'the net' so that it is highly unlikely that any reduced final download will be due to server overload and will thus be more likely to demonstrate poor download speeds from the ISP?

3) If single thread tests give a consistent lower speed than multi, does this imply a 'local' problem or just that a different server (source) should be tried?

If it's relevant, I'm using a standard Win 10 desktop (or laptop) with Firefox (or Edge) with a nominal contractual download of £30Mbs (which in practice is varying from near 3 to 29) supplied over a radio network operated by a local ISP (ie not cable or fibre). Similar results arise whichever equipment I use, including the laptop connected direct from the radio antennae (ie by-passing the router and therefore wifi.)

Any help or advice on the above will be very much appreciated.

Is there a daily or weekly pattern to the bandwidth slowdowns - peak times (ie evenings) or is it inconsistent?

It could be an oversubscribed backhaul from the nearest radio node. Have you discussed your concerns with the provider?

Certain fixed line broadband providers now offer a minimum bandwidth guarantee on new connections - typically set at around 50% of the “headline” rate. Does your WISP offer anything similar? If you are only getting 10% of the headline speed, then I’d be pushing them for a resolution - that’s not really good enough.