The "unable to select accurately by click" issue is very obvious esp when a character is first imported from photoshop where parts are square images with lots of empty space before converting them into mesh. The characters I am working on is generally side view. You can imagine there are lots of overlapping between the parts. e.g. near/far arms and stuff on arms and main body, and stuff on the body, etc. Most of the times, parts are blocked by other parts (even below them).Selecting on canvas is always not easy until I make them meshes and set processed part unselectable. So being able to select what I actually see under mouse click accurately is a great help. For now I can just search through the Hierarchy panel to find what I need which is not intuitive enough.
Method 2
Photoshop has another way to help this issue too. When R-Click on the canvas, the popup menu show up with all the layers with visible pixel under the mouse location. This is another way to allow user to make selection. But of course in Spine, the R-Click is used for other things. Similar function can be done by checking which parts are under the cursor and show them as a list at the corner of the canvas. Example:
1.Body
2.ArmUpperL
3.Hair
User can then just press the 1-9 key on the keyboard to select one of the parts listed.
Method 3
Another way I could think of to improve the selection is to detect whats under mouse cursor and allow quick circular switching. e.g. just a single key press like PgUp/PgDn. Let's say when A, B, C, D parts are detected overlapping under the mouse cursor, PgUp could be programmed to switch selection between these parts so I could just keep pressing PgUp until the the desired part are selected. It would still be much faster than finding that part from the Hierarchy even if quick search filter is implemented in the future.