Fix bugs and inconveniences around blunder thresholds table

[kaorahi]

See the commit logs for details. As this PR includes various changes, it will be better to test it for a while before merging.

[hope366]

Thank you for the detailed correction.
I didn't really understand how to use blunder color, and I shied away from it, but in the process of applying and validating your PR, I came to understand how interesting this feature is.
Now, I want to check the following two items

• Fix IndexOutOfBoundsException when one try to empty blunder thresholds
• Quick & dirty patch for NumberFormatException in blunder thresholds

Before application : If you leave a threshold blank or enter a double-byte value, the next time you check, the value will not be displayed, but the value that was previously set will be displayed.

After applying: If you leave the threshold blank or enter a double-byte value, the next time you check, it will be displayed as "-777.0".

Is it correct that the above PR is intended to be this kind of content?

When I enter a threshold in full-width characters, it looks like it's accepted at the time, but hitting the space bar doesn't turn the analysis on or off. This may confuse users who are not familiar with computers.
I think it would be easier to understand if an error message is displayed to let the user know that it is an incorrect input when you enter a threshold in full-width characters and click "ok".

[kaorahi]

Thanks for testing. Yes, I intend to convert invalid thresholds into a nonsense value -777.0.

Your proposal is desirable of course. But it needs additional coding that I'm not willing to do. So I made this "Quick & dirty patch". Without this patch, even one mistake make you lose ALL the changes in the table when you click the OK button.

Would you post a new issue for your above request if it is important for you? Though I don't have enough energy now, I or someone may possibly implement it in future.

[hope366]

What I suggested is very trivial and I don't think it's important enough to raise any new issues.

By the way, I have noticed something in the course of this examination of blunder color.
I experimented with the official release of Lizzie-v0.7.4 with the following configuration


Black's first move is -4.9% compared to the best move, so the node should be green in color. Similarly, white's second move should be red at -14.7% and black's third move should be yellow at -5.2%.
However, any of the moves are light blue, indicating -1.0% to -3.0%.
This problem seems to have been fixed after applying Fix missing marks and color in the variation tree.

[hope366]

Mistakes were made due to lack of verification.
#800 is probably not relevant.
The key to determining whether it would work or not seemed to be the order in which the thresholds were entered.
The thresholds must be entered in order of decreasing, and if you specify -10 followed by -20, then -20 is ignored.
I mean...

In this case, orange and red are ignored.

With this setting, everything but the light blue will be ignored.

Sort blunder thresholds before recording applies, the thresholds you enter are automatically sorted in decreasing order, so it seems to work no matter what order you enter them in.

[kaorahi]

I read the document of JTable and noticed that we can simply store Double into it. So I pushed another commit for better behavior.

[hope366]

If you enter a character that is not a half-width number, the cells become red borders and you cannot enter blunder color.
This makes it easier to see that there was a problem with the input method.😄

[hope366]

It seems a bit underwhelming that the best move, or a move close to it, is represented by colorless = white. I think it's good that the best moves have the appropriate color on them.
Is there a better way to do this?

[kaorahi]

I pushed yet another commit so that you can use threshold=0.0 to specify the color for "normal" moves.

[hope366]

Thank you very much.
I wasn't very interested in the variation tree, but now I can enjoy it a lot more because of the various improvements!😄