My backgound is maths not physics if that helps explain any misunderstandings.
Sure, you can apply all sorts of metrics. I think you are right with respect to timelike and spacelike paths and if I could confirm/concretetise what you say:
Given a point in spacetime as a reference, if you travel through time but not space (not as tricky as it sounds - just wait) with respect to your reference point then this is nothing special. If you travel through space but not time with respect to your reference point you are stepping outside a spacetime cone and travelling faster than the speed of light. In conclusion no matter how you measure spacetime, timelike travel and spacelike travel are very different and this is due to the fact that there is a finite speed of light.
Sticking to Minkwoski spactime and inertial refernec frames:
Space and time are mixed in spacetime in a way that it's not easy to pull them apart, e.g. in one reference frame a separtion can be purely spatial or purely temporal but in another it's temporal and spatial.
What distinguishes spaclike and timelike seperations is that there's always a refrence frame in which a spacelike seperation is purely spatial and never a reference frame in which it is purely temporal, equally there's always a refrence frame in which a timelike sepration is purely temporal and there's never a reference frame in which it's purely spatial (to avoid confusion despite what I said early it might better to think of lightlike seperations as the limitng case of timelike seperations)
Surely multiplication by scalars still work normally?
As you've already noted that's not a property of metric spaces.
Absolutely. The popular notion that 'time is the fourth dimension' is positively Newtonian. However I think physcists used to regard time as a dimension (a component of a cartesian product) multiplied by the square root of -1. I'm not clear on why they dropped this.
The ntoion of a diemsnion is a little bit more specfic than the cartesian product, though it's related as handwavingly the dimension of a space is how many numbers you need to describe each element.
Using multiplying time by i in spacetime is a cheat just so we can deal with 4 diemsnional Euclidean space rather than Minkowski space, but it's a cheat that ignores important properties of Minkowksi space. For example if you continually rotate something through an axis in E^4 after a while you'll get back to the same psotion it staretd , but in Minkowski space if you do the equiavlent it'll never get back to the same postion it started.
The reason that people call time the fourth dimension is taht when you define your basis vector fields over spacetime in order to create a reference frame you have 3 basis vector fields made up of spaceliek vector and one basis vector field made up of timelike vectors. What I'd say is though that Minkowski space is four diemsnioanl before you define your basis vector fields and it's only really when you wnat to get numbers out of the equations that you have to define a basis. Also as I said earlier soemthing that is purely temporal in one refernce frame can be a mixture of temporal and spatial in another.