The only place where Scrivener generates LaTeX code is with its “General Non-Fiction (LaTeX)” project template, which isn’t in the beta yet. But in that case you have complete control over how the compile format works; there are no special features in Scrivener that are aware of LaTeX as a syntax or system. To see what Scrivener itself produces from your project, use the plain “MultiMarkdown” option and examine the text file. Do you see the codes you refer to anywhere in here? That’s a good general rule of thumb for figuring out which set of documentation to look at.
What you are looking at here is MultiMarkdown output, the same sort you would get if you typed in your document into a .txt file and use mmd2tex on the command-line (or however that works on Windows). If you would like to customise how MMD works, you might have better luck on the MultiMarkdown discussion board.
Sorry I don’t quite follow what you mean here. To my mind, a plain-text file is what a .tex file is, just like an RTF or HTML files are plain-text, and I don’t know what a .tex file without code in it means. But hopefully you just meant what I was referring to above as a troubleshooting step: just use plain MMD from Scrivener if you don’t want LaTeX syntax in your file at all.
Ah! one thing you might be battling against: you may have been using an old version of MultiMarkdown before. We don’t mark raw LaTeX code with HTML comments anymore, comments are just comments now. A while ago I described how I make call-out boxes with LaTeX and MMD. There are some practical examples given a few posts down.