Improve memory performance for txt chat download

[jmSfernandes]

WHY
My use case includes running this cli to download twitch chats, in a constraint environment (e.g, Rapsberri Pi).

The tool is great, but for large streams (with a lot of messages) the memory footprint can be big.
for instance for:
https://www.twitch.tv/videos/1857424785
The tool takes around 510 mbs of memory (see images).

I was looking at the code, and I found out some points in which we could optimize.
For instance, emotes, and Badges information are not used for txt files.
And creating these instances takes a lot of memory.
With the proposed changes I was able to reduce the memory usage significantly, to around: 360mbs for that particular case.

I also added a quiet mode (-q) option to the CLI, since some users might not be interested in keeping track of the progress (for instance my case, were the process is automated and I'm not looking at the progress), and this also lowers the memory consumption.

Improve memory performance.

HOW

• Add quiet mode option (-q).
• Improve Datetime parsing.
• Added conditions to verify what is the Ouput File format when converting comments.
• Changed Comment's message body creation to use a StringBuilder to improve memory performance.
• Change progress percentages List to an int array to be allocated in the stack and initialized at 0.
• Change DownloadSection to return a List of Comments objects, and made the comments unification and sorting after all tasks are finished to avoid concurrency issues,

Note:
Hi this is my first attempt to contribute to this project, so I'm sorry if I overlooked something.

[ScrubN]

Improve Datetime parsing.

I can't tell exactly what this is referencing, could you help point it out?

Added conditions to verify what is the Ouput File format when converting comments.

I guess this is alright, though several people have requested mock chat badges in text chats so this may be partially undone in the future

Changed Comment's message body creation to use a StringBuilder to improve memory performance.

I spotted this myself the other day and it would have been implemented anyways by #752

Change progress percentages List to an int array to be allocated in the stack and initialized at 0.

Unless the stackalloc keyword is used, int[]s are still allocated on the heap, so this really only saves about 32B of overhead introduced by the List<T> class, though every bit of saved memory still counts

Thanks for the changes, though if it's alright I would like to push a few minor changes to fix some formatting or some missed opportunities for further optimization