Say you have a 100 x 20 x 1 mm cantilever beam and you are bending it in the thin direction. There will be tension on the top and compression on the bottom. If you mesh that with solid elements, you want at least 4 elements through the thickness to transition from an element that sees tension to an element that sees compression. If the geometry is not some simple brick, but a complex thin walled solid, it can be very difficult to get a mesh that puts 4 layers of elements through the thickness. You saw that the Tet mesh you had only had 1 element through the thickness. That means a single element has to represent tension on one side and compression on the other side. It doesn't do a good job at that.

If you make a midsurface from that solid, you will have a surface that is 100 x 20 mm and you put a SHELL mesh on that. You can assign the SHELL elements the 1 mm thickness. The formulation of the SHELL element includes bending equations, so a single element is designed for bending. This is much more efficient than the solid elements. If you extract the midsurface of that complex thin walled solid, it will be easy to mesh that surface with shell elements.