are they contract bound to use them?
I suspect it varies with the way the various teams are supported by the manufacturer. If they are sponsored by them, and get their kit for free, they might be. However quite a few of the teams buy their kit, in which case (price incentives aside) they might have a free choice.
The riders will probably have a preference but one factor that may be important (in ways that are not clear) is how the maintenance is carried out, and if the mechanics are short-handed or not. Keeping a whole team's worth of bikes in working Bowden cables is no small undertaking and some mechanics may find it quicker to play 'parts change bingo' with Di2 or EPS systems instead.
There are also questions of failure modes and how any in-service problems are perceived by the riders and the team management. Personally I think it is almost inevitable that electrical systems of this sort will play up from time to time; for example electrical contacts only have three states; connection made, connection lost, connection intermittent. Two out of three states will give an unforgiving failure mode, and one of them may cause endless trouble before it is identified and corrected. [By contrast Bowden cables usually carry on working (badly) for quite a while before they pack in altogether, which is arguably a more benign failure mode.]
When a Di2 system goes wrong this could be blamed on
- the mechanic
- the rider (doing something weird with it)
- the system
In this day and age if a system is found to be unreliable a common approach is simply to replace all the parts in it more regularly. This approach is predicated on the assumption that the parts wear out or otherwise deteriorate in some way, and is no substitute for improving other aspects of the system (since at best it will make failures less frequent, rather than eliminate them), but it may seem like a good choice if you are "in that particular hole".
It is an interesting conundrum; if the Di2 shifts are seen to be even fractionally quicker or better or less tiring for the rider, this would be seen as 'a marginal gain' which would have to be balanced against the risks and consequences of system failure. This kind of risk/consequence judgement is notoriously difficult to do well, and how any issues are perceived and managed within a group of people can affect everything.
I feel sorry for the mechanics, actually. If you assemble a Di2 system (as per factory specifications) and test it thoroughly, you still can't be absolutely
sure it will carry on working for five weeks, five days or even five minutes, and when it fails it can't always be known exactly what has gone wrong. I suspect that most experienced mechanics would feel more confident that a mechanical system would work well in the short term, and/or fail in a more graceful fashion.