Implemented "skeleton" SUBs

The first step towards supporting SUBS/FUNCTIONS is complete. The "skeleton" SUBs have no shared data and accept no parameters, however they behave properly in any other possible way. They indwell and create/free data (including arrays) appropriately for the scope of the routine. QB64 currently ignores DECLARE statements and instead scans for SUBs/FUNCTIONs in a prepass. Of course, "skeleton" subs are not useful, but they are the first step in QB64 fully supporting SUBs/FUNCTIONs.

RE: Bug discovered in QB64 Demo #5 (by me!)
Any strings not created as part of an array will be declared in the wrong memory region, the DBLOCK emulated conventional memory region. This means you'll get an out of memory error in the following code:
a$=SPACE$(65536)

RE: Recent SPAM
Very annoying. Unfortunately I don't have the privilages to delete such posts. So for now I'm just ignoring them.

RE: Frequent references to FB
Also usually annoying. If you think the goals of QB64 are pointless don't bother complaining because the goals aren't going to change. Most comparisons made fall under the catagory of "cheap" point scoring for the purpose of pointing out that FB is "better". Of course, these usually ignore the most important part of QB64, compatibility. I word which, in my opinion, you cannot put a price on. Those posts claiming FB is more compatible with QBASIC than QB64 are stupid for obvious reasons. Sure, it you want to program in a 32/64-bit most like QBASIC language NOW go use FB, but 6 months from now that language will be QB64.