Fee

My understanding of the Full Screen API is that it only works during user interaction, so it cannot be used maliciously. As per:
https://stackoverflow.com/questions/19355370/how-to-open-a-web-page-automatically-in-full-screen-mode
Maybe some context would help us understand why you feel the website needs to go full screen, and why you think interaction will interrupt this process? Full screen has a very narrow usage, for example, as used by YouTube, when I watch a video I don't want to interact with my browser for a time. So, I temporarily hide that functionality, if I choose! Auto full screen on YouTube would be annoying - something to consider. For still images, I wouldn't imagine wanting to view them for longer than a couple of seconds before I continue to interact with the browser. For that interaction, I need access to the browser's features I do not want them hidden from me. When you add-in what could be considered intrusive interaction (full screen, audio etc.) always allow the user to opt-in, rather than opt-out. 
 

Use preloadJS to load a bunch of images, then EaselJS to provide the drag-and-drop functionality: 
https://createjs.com/
The following tutorial will show you how to achieve drag-and-drop with EaselJS: 
https://www.createjs.com/tutorials/Mouse Interaction/
Sending you the end design is slightly more difficult... You could track everything that happens on the canvas in a JavaScript Object. When an image gets added, add it to the object, when a user drags it, update the X and Y coordinates. When the user hits send you'd simply need to JSON.stringify the object and either save it to a database or send it to yourself somehow. Although saving in a database is probably the optimal solution as clients would be able to revisit their design at a later date. When you'd like to view a client design, you would need to read the JSON string back in, parse it and then loop through the JavaScript Object to rebuild the canvas. 
Alternatively, using the following two libraries, you could export the canvas as a PDF: 
https://html2canvas.hertzen.com/
https://parall.ax/products/jspdf
I've made a similar app which is used by children to build an interactive Life Story Book (https://en.wikipedia.org/wiki/Life_story_work) and I can tell you from experience that what you're suggesting will be a massive undertaking in terms of development, before it is anywhere near useful. That said, the above will allow you to make a proof of concept that allows clients to drag some images around a canvas and save the result. 

I really like the logo. 
A couple of things I picked up on, just had a quick look. 
On the home page I think there could be a good deal more contrast between the navigation/logo and the background. I'd also probably prefer a sticky navigation. 
I think the "Five stars on Trust Pilot" section should link to trust pilot for authenticity purposes. 
With regard to CSS Google prefers one file. If you minify it, even better. Some people will have two separate CSS files; The one 'main' CSS file contains layout-essential CSS -  everything else gets loaded after the fold in a separate CSS file using a <noscript> tag. This in theory should speed up page load somewhat, although I've never tested this personally. I personally stick to one CSS file, minified. 
 

I think SVG would be the way to go. Using the border-radius property would be a possibility, but since as far as I know it isn't possible to create a double curve in CSS, you would need multiple pseudo elements to achieve this. You might even be able to do this using a radial-gradient as a background image, but again I would suspect you would need pseudo elements. It's actually quite a tough shape to create, I had a go using the two CSS methods I mentioned above and couldn't get that close. 
The background gradient on the other hand is no problem. Use a linear-gradient property on a background image, as seen here: 
Code Pen
Linear-gradient is supported by most if not all browser I believe. 
 

When building out a UI I just forget about individual devices like mobile, tablet etc.  (At least in terms of layout, not talking about gestures / accessibility here).
Essentially the layout should look great at any reasonable screen size, I tend to use mobile-first media queries (My framework is setup this way) but designing one component at a time, make sure it works well in all situations before moving onto the next.  We should always be thinking about things in terms of components rather than pages.  To take that further Atomic design is a way of composing components into larger groupings which can form part of or a whole page.

If you haven't invested much time yet into Angular then ditch it and look at something like React