Some points require clarification or corrections:
Question 9:
There is a slight inconsistency in terminology.
You rightfully it says that
Proof of Stake (and Proof of Work) aren’t consensus mechanisms - they are sybil-control mechanisms.
but two paragraphs later it says
There are currently multiple projects that employ the Proof of Stake consensus mechanism.
Please make clear when you speak of Casper FFG and when of Casper CBC or whether you in general talk about FFG.
Based on my knowledge about Casper CBC (might not be true for FFG) this example
For example, if a validator proposed a different version of the blockchain versus what the remaining honest nodes have reached consensus on,..
doesn't describe an equivocation as long as you are not sending contradicting messages.
Question 10:
Users are able to run multiple validators but their computing cost will increase as well (it should be noted that there is a cost to running a validator in Casper Proof of Stake including the cost of computational resources, the cost of high availability and devops, and the opportunity cost and risk of locking value in the stacking system when that wealth could be used elsewhere).
In general this is true but can be misleading for some cases. Having n staking slots does not necessarily require you to bring up n times the computational resources. Depending on validator rotation you could have one machine doing the signing for multiple slots.
If an attacker gained 51% of all Ether they could attempt to attack the network but mechanisms in place make it likely that they will still be slashed. At current Ether price, this attacker would be risking $4,300,000,000 to do so and if caught could lose it all. The cost to aquire 51% of the Bitcoin hash rate is much less.
An attack analysis should be based on the staked amount not total supply. Attacks scenarios against liveness and safety require their own analysis and have different attack thresholds - not necessarily 51%. Obv. attacking costs should also based on the ratio that is required for these attacks based on the total amount staked not on the total supply.
It should be noted that
a) acquiring a large portion of the total supply will increase demand and therefore increase the price and thus attacking costs and
b) unlike in PoW you can't start the next attack after the first unsuccessful attack because you will be slashed for the first one.
Question 15:
Yes, Ethereum's byzantine fault tolerant consensus prevents against many consensus based attacks such as the well known 51% attack, ...
Please explain, how are 51% attacks prevented?
Question 15 + 23:
Additionally, smart contract coding principles and standards are growing by the day to ensure better-quality contracts that operate as intended. Formal verification of contracts is becoming a norm as well as multiple audits before release to production.
Not only are we far away that formal verified smart contracts are becoming a norm you can argue that the EVM makes it almost impossible to formally verify a smart contract against the EVM execution.
Question 20:
The off-shore exchange BitMex, incorporated in the Seychelles, has engaged in market manipulation through its proprietary trading desk that has direct advantages over customer traders.
Reference please.