Remove markdown parsing from frontend

[siddhantk232]

marked was used to parse markdown
for each text node in the browser which was making things very slow.

This commit moves markdown parsing to the backend using the comrak
crate.

Some cleanups may be required that removes the marked.js files
completely from the codebase but it'll happen later once I get current
changes reviewed.

ftd code that was used to test this

-- ftd.text: 

Hello world `hellow` [link](https://google.com/)

\```rust
fn hello () -> {
    

    println!("works?");
    println!("works?");
    println!("works?");
    println!("works?");
    println!("works?");
    println!("works?");



}
\```

-- ftd.code:
lang: rust

#[derive(Debug)]
pub struct SetProperty {
    pub kind: PropertyKind,
    pub value: SetPropertyValue,
    pub element_name: String,
    pub inherited: String,
}

#[derive(Debug)]
pub enum SetPropertyValue {
    Reference(String),
    Value(fastn_js::Value),
    Formula(fastn_js::Formula),
    Clone(String),
}

impl fastn_js::SetPropertyValue {
    pub fn to_js(&self) -> String {
        self.to_js_with_element_name(&None)
    }
}

I couldn't see any visual changes before and after this commit. Attaching screenshots anyway:

(before this commit)

(after this commit)

[siddhantk232]

I expect the tests to fail but I'm not too worried about that right now. I want either @amitu or @Arpita-Jaiswal (or both) to first review these changes carefully as I'm not too confident about this.

[amitu]

@siddhantk232 how have you tested it so far? Is fastn.com, fifthtry, ide etc working with it?

[amitu]

Can you also create some stress test / benchmark files, eg with 10K ftd.text nodes (with slightly different content; generated using Python) and do a before and after (backend and frontend timing) to see if it indeed made any improvements or is it a performance regression?

[siddhantk232]

@siddhantk232 how have you tested it so far? Is fastn.com, fifthtry, ide etc working with it?

I tested it with the ftd snippet mentioned in the PR description which worked fine. fastn.com does have some visual differences (ftd code is not highlighted and other code is not indented properly with this change). I didn't see any noticeable performance gains either:

fastn 0.4.76 lighthouse report on fastn.com/web-component/ ran locally with fastn --trace serve --offline

Same webpage but with current build of fastn that includes this change

@amitu

[siddhantk232]

I don't think moving the code highlight part (prism.js) to the backend would make this any better. We should probably not go ahead with this for now.