Your assuming it's cost effective to turn off your desalination plant for 1/2 the day. It's not. Further, oil is a tiny fraction of total electrical power. It just used for vary expensive peaking power. Or in a small number of countries that heavily subsidize oil.
Consider those little gas generators people buy cost ~1$ / kwh. At utility scale it can be worth ~40c/w to cover the tip of the peak of power demand to prevent brownouts, but base load power is far cheaper and your base load power is still operating during peak demand reducing average costs.
This is where it starts to get to the point where I don't understand the economy of energy enough to really judge. My understanding is that with some desalination methods, energy costs represent over 50% of total cost which suggests that irregular production shouldn't be too much of a problem if energy costs are that different.
Let's say you could generate your water for 100$. With 50$ from building the plant and 50$ from energy. Now, if you operate it half the time you need to build 2 power-plants at 50$ a pop = 100$ so if you pay even 1cent for energy your worse off.
That's assuming energy is the only marginal/variable cost. But, I guess we are playing with contrived examples either way.
Anyway, if inconsistency is an inherent cost of solar it needs to be treated as such in comparisons. Even if storage gets "solved" it won't be free. There will be energy costs to it and other costs. To have consistent capacity of X, you will need a peak capacity of several times X. If peak grid usage is completely covered by solar, there will be wasted capacity at the coal plants at this times.
Storage an issue based on the % of solar adoption. At say 1% solar you don't need any storage and it's cost effective in most countries today. At say 80%+ you’re going to need a lot of storage.
Worst case is basically 100% solar which is how many home systems are designed. At which point storage basically doubles your costs now and it only gets worse as solar gets cheaper. Granted there are low cost utility scale solutions, but PV is still at low enough levels in most areas that it's cheaper to simply store nothing and vary occasionally let excess power go to waste.
PS: The important thing to note is solar is cheap enough that even without any subsides the market is going to stick with a solar component anyway simply because that's what's cost effective. And solar is only getting cheaper so that % is going to slowly rise over time without any intervention. The issue is people want solar to make a large percentage of the grid power and that's where there are problems.
Consider those little gas generators people buy cost ~1$ / kwh. At utility scale it can be worth ~40c/w to cover the tip of the peak of power demand to prevent brownouts, but base load power is far cheaper and your base load power is still operating during peak demand reducing average costs.