Refactor Vault using canvas instead of only HTML/CSS

[mneunomne]

Here are considerations for the implementation:

PROS:

• Zoom in/out already part of the libraries
• Better performance since there is only 1 html element being manipulated <canvas></canvas>.
• Possible to implement other visualization models if we want to (network, 3d)
• using collision and force it is easier to add/remove the elements on-the-go

CONS:

• More work to make the gifs animate, but possible (using libraries such as gifuct-js to parse them)
• More work to implement text segments inside the canvas
• More work to implement links
• No copy/paste

CONSIDER:

• D3 rendering performance vs canvas