Bidirectional might give some extra overtaking opportunities, but only if there are crossovers existing or provided new in the right places. Under ETCS, bi-di still has to be consciously designed into the interlocking, although with only additional block markers instead of bang road signals, the incremental cost might be smaller than with signals. it might be possible to retain some signals in the normal operating direction on each track, with blocks further subdivided around station areas to allow better closing up through platforms. This would retain capability for small numbers of non ETCS equipped trains at off peak times perhaps or to provide a 'fail back' to keep trains moving at least in the event of radio or some other system failure; this is how the Thameslink core is configured. The reverse direction moves might only be configured with ETCS, hence not requiring the bang road signals which often require large and complex signalling structures to provide suitable sighting. Thus an express train fast modern made up of modern ETCS stock could switch over to the 'wrong' line where expedient to overtake an older stopping train that was not ETCS fitted.
You make most valid point which many proponents of the benefits of ETCS ignore. Each and EVERY route, be it ETCS or conventional requires the same specific interlocking logic and route setting request logic from ALL route setting sources together with whatever protection is specified for overrun's beyond the end of movement authority. The design and associate testing of all of this is where the time and expense is spent. The cost of system equipment to provide is a sunk cost as in any one area the cost of equipment for
additional routes is negligible compared to the time to design and test the routes as that equipment will already be provided for the core signalling function. The more routes, the more complex the design, the more protracted the safety verification and validation process and room for error. Recent history tells us that in spades. Signal sighting constraints are real and problematic. The solutions however when measured in cost terms normally pale into insignificance.
As for the practicalities of operation. Through operations like Thameslink core and Crossrail where trains keep moving ahead (in theory) fine, but what is the point in context with the above of closing up trains on the approach to terminal stations where platform occupancy and vacation time will dictate how quickly and therefore how many trains can be timetabled. In context closing trains up a few hundred meters will make little difference. They'll just come to a stand nearer their destination and sit their just the same. Which is why imho the idea that ETCS will be the holy grail of extra capacity is flawed unless you solve the end destination problem. Trains will merely speed along closer together en route and then wait for a terminal platform further away than they do now as they join the back of the queue. I'm also deeply sceptical that stopping your train on a complex approach at the allotted block point without intervention and without a
clearly visible position marked where clearance points are numerous and critical will be less than straightforward.
Hence CR2 is seen as as one expensive but operationally practical solution for one location.
