The Dominion comparison is very apt. Most TCGs have minimum deck sizes but no maximums for this very reason. The power of the average card you have access to is the limiting factor. In most games of Dominion buying copper dilutes your deck, even though it is money.
Combining a point buy with a minimum deck size would "solve" that problem, but unless you have 0-point cards it can be very annoying to build a deck exactly right.
Card copy limits are also make thinking about your deck easier. If you judge a card to be very powerful or essential to your strategy you automatically include the maximum number of them. In a point buy system do you include the max and then fill with 0-costs?
This leads into the next problem - variety. Decks become boring to play when they're guaranteed to do the same thing every time they play. If your deck is just 50 copies of the same card it won't be much fun for anyone.* Card count limits combined with deck size minimum solve this as well. So unless you find another solution you'd still need these limits.
I think point-buy is a poor tool to solve deck construction problems. It might be a good tool for solving card-playing costs, if you don't want to have such a system during the turns of the game. Basically, if you remove casting(playing) costs from cards and replace that with points during construction you can change the play mechanics to "play one card per turn" or "play any number of cards" for free. This creates pacing problems, though. If you draw the most powerful card at the start of the game there won't be a nice dramatic build up to it. Also you lose on the fun anticipation of trying to build up the resources to play it while you stare at it in your hand turn after turn.
TCG design is hard.
*Plague Rats work out because they are an exception to the rule - they surprise you and amuse you for a short time but you don't want the whole meta to be like that.