I'd go with eclectic, though I don't always come as quickly to the same conclusion as I'm as tight-fisted as the next man.
It's easy to look at a near doubling of price and see that as 'exorbitant', easy to forget that in total terms of a project or the value a project has that's usually only a smaller increase, and especially if one casually discounts time and effort as being "free" which we all tend to, after all it's usually a hobby we do for fun.
It really comes from where one gets one's hobby fun from; if it's the fun of downright determination to get something working on a particular chip come what may, by all means have that fun, go for it, commit the time and effort, enjoy that challenge for what it is.
If satisfaction and fun comes in getting the project working, as you want, with fewer up-hill battles it's often worthwhile spending a little more to save something somewhere else ( which is usually time and effort ).
There's always a balance, and I think we're all in the game of as cheap as possible in most cases and enjoy the challenge, so there's no hard and fast rules. There's always a satisfaction in getting something to work on the cheapest or smallest, the joy of beating, "they said it couldn't be done", but there's no shame in taking the easier way out and spending money to get to where you want to go to.
In terms of optimising PICAXE code when up against limits. In my experience, saving 10% is probably reasonably achievable, 20% possibly, beyond that it may well require major redesign and a lot of effort and even dropping some functionality. Plus most optimisation will take away the aesthetics of the original code, make it harder to understand and to modify.
If at the limit and just needing a bit more then the project is done it may be worthwhile, but if near the limit and there's a lot more you'd like to add, it's probably easier to 'suck it up' and accept that a more capable PICAXE will likely be the best way forward.