Maybe I'm being cynical, but if you have been following the political rows and finger pointing between Westminster and Cardiff Bay over this new franchise and the rail system in Wales in general over the last 2 - 3 years, esp the rows over funding, you'll forgive me for not being optimistic about cooperation on infrastructure that will be owned by 2 different governments (as the interface between the Valleys and the mainline networks at Cardiff Central will be) who can barely agree on anything. As has already been mentioned, there' bound to be disagreements between the two 'sides' (because that's what they are) over liability.
But how would this be any different to a Labour-run council fighting the UK government? This won't be the only time there's a risk interface between NR and locally-owned infrastructure. Presumably, the laws or regulations or orders which sort out the ownership split would include a mechanism to work it out.
Even if there are disputes over payment, it doesn't stop the immediate need for a fix. If one or other of the sets of lines in Cardiff is blocked, the economic cost of non-action massively outweighs the possible cost of having to pay when you're not meant to. If something went wrong, it would need to be fixed and then payment sorted out afterwards. The random possibility of expensive catastrophe is the reason for the insurance industry to exist. The accountants would not be happy if a nominal insurance fee were not included in the books for the South Wales Metro (whether a payment to a commercial insurance company, or a charge covering the cost of effective self-insurance).
The eye-watering cost overruns on the Sheffield tram-train project and the fact that it hasn't even entered full service yet, when it was supposed to be the pilot project for judging whether to roll it out elsewhere, doesn't fill me with hope.
The idea of light-rail vehicles being allowed to run onto the NR network alongside arbitrary freight and heavy passenger services has been around since 2002 in Sunderland. The technical standards already exist for the vehicles, so building them to match would be easy. We already have 7 (well, 4) actual tram-trains about in Britain today, even if they don't currently do the train bit.
The South Wales Metro works would have to be phased in any case. Getting the shiny new trains/trams/tram-trains/train-trams in first for the photo ops is a fairly reasonable first step. The infrastructure can be modified and street running sections added incrementally over the years to come. I really wouldn't be surprised if this happened. It looks like there really wouldn't be a lot of 'wasted' wiring if the core Cardiff local routes were wired (with a mixture of 750V DC and 25kV AC) before the street running sections opened. They might want to convert the Penarth line to LR operation with street running through Grangetown, but in the meantime extending the GW 25kV AC down to Cogan isn't going to be a bad idea.
I can’t see the benefit of having trams on the streets of Cardiff over buses - apart from the fact that they will not have diesel fumes. If trams are descending onto the streets of central Cardiff from all over SE Wales, surely chaos will ensue as they are not able to overtake each other - unlike buses? Just stand on Westgate Street and observe how many buses simply overtake those waiting at bus stops. Can you imagine it if these were trams - all trapped behind each other? Buses can also switch routes in the event of a problem/sporting event.
Trams don't overtake one another, and don't really need to. Buses have to because they are inefficient at loading and unloading. A 50m single unit set will have 6-8 doors on each side, all with step-free access to the platform. This means dwell times can be very low even with large turnover of passengers. Dwell times are low enough that it's perfectly viable for trams to just follow one another along a set track, as the dwell times aren't long enough to reduce capacity on the line.
Also, when you're designing a tram network, you don't use the street running ability to just follow existing traffic through the city and call it a day. City centre traffic flows often need a bit of a re-jig to minimise interaction with the trams. When designed, they aim for traffic separation at best (e.g. running through a pedestrian area or along the side or median of a road), then shared with public transport only (e.g. Princes Street in Edinburgh - buses/taxis/trams only), and then completely shared traffic as a last resort.
Modern tram networks are trying to build an underground-style Metro without needing to dig, rather than just putting buses on rails. Trams provide a happy medium between a high-capacity metro line (e.g. the Victoria line) and a normal bus service. When cars came about, many cities thought that their old tram networks would slow down cars in city centres, so spent huge amounts of money building fairly low-capacity underground metro or pre-metro systems. Now that we see cars move no faster through cities than horse and cart, and that cars really don't mix with public spaces, it's clear that these moves were a mistake. When you need a medium-capacity metro solution, the go-to system is now some sort of modern street-running tram. The only real reason to tunnel is if the urban environment simply doesn't allow trams due to age and complication (which might be the case in Cambridge).