- Have devastating effects on the fee market
Since there's more space in each block, this space is now less valuable, fees might not be zero, but they'll be lead down to the minimum and competitiveness for fast transactions will be dead.
I don't know how it would change the fee market.. There hasn't been any scarcity in blocks before so fees would probably stay quite similar to what they are today. If adoption continues to increase, the total of collected fees will be higher as there would be more transactions per block.
Wrong. Some scarcity in blocks, and some disparity in the fees paid is essential to maintain a healthy economic relationship between the miners and the users. More tx volume just recreates today's dynamics on a larger scale. As the block reward subsidy diminishes, the extent to which that subsidy was distorting the fee market will become increasingly more apparent. When the block reward approaches zero, having a functional fee market in place is the only way of incentivising miners to mine at all.
Treating fees as "one-fee-fits-all" doesn't take account of the economic reality. Some transactions must confirm ASAP, for a wide variety of reasons. Others can be afforded hours, days or perhaps even longer than that.
- Bring nodes offline and make pools/miners less reliable
The change would increase bandwidth consumption as well as HD space required to run a full node. There are many nodes that run on non-optimized hosts, if they suddenly start sucking up more of the bandwidth and HD space it's safe to assume that many will go offline. Also, pools running on weak networks would be more prone to orphaned blocks because of the blocks being harder to propagate.
One possibility is that increase in adoption and advancement of technology will balance that or even make number of nodes increase. We can't know for sure.
What I know is that letting blocks to saturate will make the fees rise rise and will make bitcoin less inclusive. It will be less appealing/useless for many markets/use cases thus stiffle the growth. Everyday users will find it less useful and investors don't want to invest in niche product with no growth potential.
Again, you're speaking as if every fee will be pushed up to the same level; there is, and always will be, a range of fees that can be usefully applied in order to target the confirmation time the user needs for a given transaction. A functional fee market will make instant confirms more expensive; they should be, because instant confirms are in high demand.
Bitcoin Core wallet has only recently introduced the "smart fee" feature that helps the user choose the right fee for the confirmation time they want (there is an API for other software to use the smart fee feature too). A software feature that was written by the designer of BIP101, no less.