Please understand that we are a small company with limited resources and time. We do our best to meet the requests of our customers however we get feature requests daily from the thousands of customers we have. We have to prioritize them based on various criteria, with bug fixes being the highest priority. While we would love to implement every request that comes our way, we simply don't have the manpower to do so.
Back to the issue though. Loading of source is done mostly in a worker thread with a callback to the main thread to store the results. The actual parsing is done by the semantic parser service which runs in a worker thread if it is enabled. Do you have that enabled? If not, it will run in the main UI thread, which is bad. Once it finishes a document, it calls back to the main thread to merge the data into the SourceProjectContent but that should go very fast.
If you are sure it is still slow, then perhaps you could put together a small project that shows the slowness of the startup. This way we can debug to see if there are any immediate suggestions we can offer. And if not, perhaps we can continue discussion on how to implement cached source content. There are some issues to consider related to that which would need to be discussed.