2. They chose SPHINCS+ which have 49,856 bytes. Without also proposing to increase block size, it would massively reduce TPS (transaction per second).
On the website
https://quantum-resistant-bitcoin.bitcoin.foundation they mention SegWit v3+:
SegWit v3+ witness structures efficiently accommodate large signatures without blockchain bloat, while priority mempool treatment for QR transactions during transition phases creates economic incentives for adoption.
They don't explain how it prevent bloat. But it works just like SegWit, it means SPHINCS+ signature data still have size 49,856 bytes while have less size if you use unit weight or virtual byte which doesn't prevent blockchain bloat.
I don't see why it cannot be increased to the extreme, let's say to 2100000000 bytes (2.1 GB). I don't think there will be ever so much transactions in the block to take up all that huge space.
No, it would open gateway to massively bloat Bitcoin blockchain at relative low cost. Ordinals hype in past have proven this. There are other technical issue such as time to verify and propagate block with such massive size.