What is the testing pyramid and how does it guide test distribution?

Imagine building a toy tower with blocks. The bottom layer is made of many small, cheap blocks that are easy to stack quickly. The middle layer has fewer blocks that are a bit more expensive and take longer to place. The top layer has just a few big blocks that are the most expensive and take the longest to balance. This is like testing, where you want lots of simple tests at the bottom and only a few complex ones at the top.