The question of how much battery is a tough one. If you are off grid, you do need to use a lot of battery. Two day of battery only is probably a good place to start. On grid can be so much more cost effective. When I did my initial design for adding the storage battery, I thought 10 KWH would be plenty. I ended up with nearly 18 KWH because I found a great deal. And now I wish I had more still.
Currently batteries are still expensive and certainly cost more than the solar panels. My current project is to add 6 more panels for DC charging of my battery bank. The 6 solar panels are such a small part of the cost. I wish I had room to put up more, because the balance of system cost will not change much. Where I am, I do not have many choices. We have a lot of codes here to deal with, and there are only so many places I could put panels without running into compliance issues. So I totally agree, if you have room for a larger array, just do it. I would have put 4 more panels on my original Enphase system, but I would have needed to also change out my main breaker panel, which would not only have been more cost, it would have caused issues with getting the permit and inspections also.
Having my Solar for almost 2 years, and now the battery inverter for 6 months, I have a much better idea of my real loads. One thing I stumble on, is that I really like the idea of having BOTH microinverters on some panels, and a DC charge controller on the rest. A system like a Sol-Ark or Outback Skybox does sort of work like that, but being all in one, it is still a common failure point. Here is my thinking.
Any power that you use while the sun is shining is going to be most efficient coming from microinverters. If they do make more power than you need, the extra can either back feed the grid, or be fed back into the battery through the battery based inverter.
Charging batteries will be the most efficient from a good MPPT DC charge controller. But if you are also feeding an inverter at the same time, it is less efficient at making the AC power than the microinverters. The battery power should only be used when the sun goes down and the need to use the battery power. So the DC array is sized for the overnight loads, and the AC array is sized for the daytime loads. And having a bit more than you need is not a bad thing. You can toss out power you don't need and can't use, but if you don't make enough power due to clouds etc., you are going to have to buy it from the utility, and put more fuel in the generator.
As for the battery size, I think most of us here are limited by budget more than what we would like to have. If you can afford it, it would be great to have enough battery to run everything you want to for 2 full days. If you don't have good sun, that give you time to get ready to fire up the generator. This is not practical in all cases. In the heat of summer hear, I run my A/C and I just know I need to buy some power from the grid for that. If my system runs for half of the day, that will need 12 hours at 15 amps x 240 volts = 43.2 KWH's just for the outdoor compressor and condenser fan. That would need 3 times the battery I have, and a LOT more solar panels. Without my A/C running, I use just 25 KWH a day to run everything else in my house. My Enphase solar array has 4,800 watts of panels, limited to 3,900 watts of inverter, and makes about 25-30 KWH's a day for 9 months of the year, when it is sunny. My proposed DC array will add about 2,100 watts of panels, and produce 10-15 KWH's per day to charge the battery bank. On most days, the DC array alone should be able to top up the battery.
Sizing a system to minimize the cost can be tricky. Making a system that can make enough power in a limited area can also be a problem. Battery storage is still expensive. There is no one rule that works in every case. Having plenty of room to use a lot of cheap solar panels is a very good place to start from though. If I was starting over, I would probably reverse what I have. Use 1/3 of the array as microinverters to run the day time loads, and 2/3 of the array DC coupled to charge the batteries. I look at it this way. You have less than 1/3 of the day to make all of your power. 1/3 of the time with sun, and 2/3 of the time, running off of the batteries. You might use a bit less power over night, but you also get just 5-7 hours of sun. And you need to use more lights at night. I think this is a fair trade off.