I think you've kind of answered your own question there. The amount of data needed to download to capture the market data would be quite large when compared to the rest of the entries and if you go down the route creating a different file as you suggest, well, that needs a second system to read the market data which is what we already have.
I've created a utility to read mission data from the journal file for my squad, but sometimes they have difficulty getting it to run because I suspect that occasionally the file is locked while being written to when I want to to read it. I have implemented a retry loop to get around this, but throwing a request at a server and reading the result is a lot easier and neater than reading an active file. I wish we could get all the data from simple web requests but then it is also useful having the historical data sitting on your hard drive without triggering loads a network traffic.
The fact is it is more practical to have certain data recorded locally and other data available on demand and so far, I think they have it about right.