so if you don't know what this is about: for some reason, scratch applies a rounding to 10 decimal places to the trigonometric functions sin, cos, and tan. however, you should be getting an accuracy of at least 16~17 decimal digits, since that is about how many digits a floating point number is able to store. so scratch cuts off the accuracy of these functions for no real reason. for a lot of common applications for trig functions, such as 3d engines, this slight rounding error wont matter much. however it really starts to matter in certain applications that require as much precision as possible, such as calculating high-point FFT's as accurately as possible in order to do convolutions quickly to a good degree of accuracy. in these cases, some alternate method to calculate the trig functions is often necessary to get good accuracy. what ive attempted to do here is provide a faster and less memory-using alternative to other common methods for calculating trig functions to full precision, such as taylor series and look-up tables, however, this project uses neither to achieve the same thing, meaning that it avoids having to run loops or create massive lists to be accurate. it works by doing a single newton iteration on an inverse trig function in order to calculate the remaining digits of the approximation. while not shown in the demo, there is also an accurate version for tangent inside. maybe this will be useful for someone idk
