The research is not saying that it has been spreading in the UK since September.
Strain A is the virus found in bats. That then mutated into strain B, which mutated in to strain C.
Strain B is more common than strain A in Wuhan so they have hypothesised from the earliest tests taken that the first cases may have been 500 miles away in Guandong where most of the cases were strain A or, at least, somewhere other than Wuhan.
If the virus mutated at a constant rate, it is possible that strain A first started circulating in September IN CHINA based on when strains B and C emerged. That is a very big if though.
The only cases of strain A found in the UK were the first 2 cases in York (a student and parent from China). Strain B, the type most common in the UK, was first detected in China in late December. Therefore coronavirus cannot have been widely circulating in the UK months before that.
The Oxford study that suggested 50% of the UK could have been infected and first cases could have been as early as January in the UK if assumptions about the mortality rate, R0 etc made in other models were incorrect (we can only estimate these based on observations/data). That was very much a theoretical model that was about pointing out that we could be getting the models very wrong as we are basing it on assumptions made from very little data so we need widespread testing.
Given that the hypothetical Oxford model was based on a lower R0 (how infectious the virus is), a lower mortality rate, and higher percentage of asymptomatic cases than current data is suggesting (all things that would increase the time taken to reach the current levels), and that there are no cases of strain A in the UK, it seems unlikely that that coronavirus was here before January at the earliest.