Trying to get back to your questions, the answer to all of them is basically the same:
Use what you've got, and do anything.
1) Yes, flash has been trending downward for several years. But that doesn't mean it is dead. It isn't dead, and even on present course it likely won't be dead for at least ten years. There is an enormous amount of old stuff on the Internet, and dropping flash means upsetting a lot of people.
HTML5 + JavaScript has been trending up for years. It is still a frustratingly difficult path, and in spite of companies claiming to do all they can to maintain compatibility this software needs to be continuously rewritten. We've got a long list of bugs flagged "Chrome 45" even though it is in beta, and an enormous list of bugs flagged as "Edge". Every browser does its own thing, and there are so many standards that they've each chosen different sets to be compatable with, plus their own custom options besides.
Using existing engines like Unity and Unreal means you're hitting a separate platform that is moderately stable and somewhat under your control, and both of these are going to be around for many years as well.
You don't even need to go with modern tools. You can find old legal copies of systems like Delphi from the late 90s, blow the dust off the box, and use that if you really want to. Use what you've got.
For a beginner, the choice of which one to use is less important that the choice to actually get started. Pick whatever language you want, whatever tool you want, and try building stuff. You are not in a position to build a blockbuster AAA game, but you are in a position to do something fun and creative. The popular game 2048 was a clone made over the weekend done as a learning experiment for the platform. There are many quirky little fun projects that are fun because they aren't trying to be everything.
2) For which platform to use, pick the platform you have. If you've got Android devices don't chose Apple. If you've got an iPhone and iPod don't chose Android. If you spend most of your time on a windows PC don't go for mobile or for Linux or Mac. If you spend your days on linux machines don't bother with Windows. Target yourself first. Use what you've got.
Use what you've got, don't worry about there being a wrong platform. The right platform is whatever you've got.
3) For scoping projects, a good recommendation is to try to do one thing every month. If you don't accomplish it after the month is over, don't fret. Don't keep working on it since that tends to lead to paralysis, just dump it and pick a new project of the month. Repeat with a new "project of the month" every month, and you'll quickly discover what you can do in a month and what is too big. Your first few projects will be too big, and that's okay. Next month, start with a smaller project; you know more than you did before, and the project is more achievable.
By doing a project per month many will be bad, but by the time school is over and you're ready to enter the work force for real, you'll have a great collection of small projects and demos you can show to prospective employers.
So repeating as I started, at this point don't worry about doing things wrong. Tinker, practice, use what you've got, and spend your time just doing anything. You are starting out and exploring, as long as you are moving somewhere there is no wrong direction.