Some aspects of testing are likely often combined with driver training runs. However there is also a need to get the trains into revenue earning service as soon as possible
And that is what should be happening, and what happened with one particular type of rolling stock in Germany is they had a huge list of test cases of every possible stopping pattern and PIS codes and announcements and checked everything and ran through them all over the course of the test runs. It was simply detailed test runs with a tick or cross if it worked and if not what happened. A member of staff from the manufacturer was present and overnight they made some adjustments where possible.
The next day they did another test run and saw if the problems still existed and made further tweaks until they got it right, making sure that every train in testing got the newer software each time it was changed before it went on another test run to make sure that there could be no getting mixed up between what software was on which always happens if you have a random mix of versions across many different units from a fleet it becomes a muddle and a complete mess.
When they finally got everything sorted out, they copied the software over to the rest of the fleet as it was delivered and there were no more issues because that is how sensible testing is carried out. All of the issues were resolved before everything entered passenger service This did involve some out of hours depot work in the early days of the fleet but it saved them a huge amount of time afterwards.
Personally I remain to be convinced by the way TL are managing this.
as there is only limited siding space available, and deliveries are happening very quickly.
That's precisely why it's best to get the issues fixed and tested as soon as possible otherwise you end up with the problems that have happened in the past, not just in GB, where something that could have been nipped in the bud turns into a whole fleet plagued by problems that will require even longer amount of time to fix because there has been such a backlog built up.
I believe that there are improvements that Televic can make to the PIS, certainly however at the same time the fact is that the testing regime that has been put in place by ThamesLink seems to be very poor and the fact that someone has a code that they were told to use that doesn't work, suggests that the management of the roll out is poor.