And you do realize the way the scientific method works, and what a theory is?
Theories are just explanations for facts, which are measurable. If a theory works, it is able to make predictions, which are then tested in experiments. This has been done for gravity in dozens of ways, and neither of them were able to prove it wrong. Theories can never conclusively be proven right, if you repeat an experiment a thousand times and get the same answer, there is still no way of knowing that the next try will have the same outcome. That's why you always try to falsify your theories, and when that doesn't happen in 300 years, it's probably a pretty good theory.
And even if gravity were proven (partially) wrong tomorrow, that wouldn't change the fact, that it's predictions on a regular scale are highly accurate. Any theory to replace it would therefore produce the same outcomes, except for maybe a few edge cases.
Gravity as a property of mass is also easily measurable independent from earths gravity, for example with the cavendish-experiment.
Also you might want to have another look at the workings of an accelerometer. Since you call it a chip (doesn't have to be one) you're probably referring to a MEMS-accelerometer. Those are made up of two tiny interleaving, but not touching, silicon electrodes, per axis; one is static, the other can move if an outside force acts upon it. This causes the capacitance of the system to change in proportion to the change in distance caused by the force. And simple mechanics lets you convert this change in capacitance to a force or acceleration (F = m * a and all that).
That means that since one of those tiny variable capacitors can measure the acceleration along one axis, so you put three of them together in linear independent alignment, and now you can measure acceleration in all directions, great!
But if you for example want to use your phone as a level, and place it on a surface, due to the case or camera bump or whatever it might not sit flat. So even if placed on the flattest, most level surface ever, it will show an offset, because the sensor in the phone itself is not level. THAT is what the calibration is for. It doens't change the accelerometer in any way, it just adds an offset, so you can use it reliably (which you should have done in your other video btw).
And for the nonconformist water: it doesn't on earth because gravity is usually pulling it away from the exterior, toward earths center. As an exterior usually has little way of stopping it, of course it flows away. But maybe you've seen videos of water on zero-g flights or even the ISS, where the water form a sphere, because there's no net force acting on it. Now imagine taking a ball of the same size, and putting it in the middle of the water sphere. Voila, now you have water conforming to the exterior of a sphere! Incredible!
Also, measuring the curvature of earth is highly trivial, you can do it with a stick and the sun, as it has been done since at least 240 BC (by Eratothenes). That is accurate enough to calculate the radius of earth with reasonable accuracy ( < 5%).
If you have any more questions, feel free to ask.