Let the window option choose a window range based on event-time limits in the data

[Constantino-Carreto-Romero]

Modify the -window- option to admit the following specifications if the policyvar follows staggered adoption:

• window(max): use for the estimation window the greatest event-time limits found in the dataset
• window(balanced): use for the estimation window the greatest event-time limits that allow using a balanced panel.

[Constantino-Carreto-Romero]

@jorpppp

• in https://github.com/JMSLab/xtevent/commit/480ba3c017e3e6cca0df345eeb4e6cd20bf0d33d I implemented https://github.com/JMSLab/xtevent/issues/170#issue-2172477057
• in https://github.com/JMSLab/xtevent/commit/746bf804376f06a088f40789f4f8f614e1e134a3 I added a list of examples to test the implementation. I also added the log file.
• in https://github.com/JMSLab/xtevent/commit/9892674815b799dc61a0cdc1cbf2f0f6c9915581 I updated the test file with examples for the implementations. Then, I ran the test file without problems.
• in https://github.com/JMSLab/xtevent/commit/70cc357f550dafee1196069a158522b92a697a6d I updated the window's description in the help file and added an example.

[jorpppp]

Thank you @Constantino-Carreto-Romero. I left some comments on https://github.com/JMSLab/xtevent/commit/480ba3c017e3e6cca0df345eeb4e6cd20bf0d33d to be addressed. Please let me know when you review them and I'll take another look.

[jorpppp]

@Constantino-Carreto-Romero can you also check the documentation to possible make the explanation of window clearer? Users may not understand why if they have event times -5 to 5, they cannot use window(-5 5). This is because of the endpoints: we should make this clearer.

[Constantino-Carreto-Romero]

Thank you @Constantino-Carreto-Romero. I left some comments on 480ba3c to be addressed. Please let me know when you review them and I'll take another look.

@jorpppp thanks

• in that same commit https://github.com/JMSLab/xtevent/commit/480ba3c017e3e6cca0df345eeb4e6cd20bf0d33d I addressed each comment.

• in https://github.com/JMSLab/xtevent/commit/7c67683a6587f0bb0f8f715fa52b989b63c6dd9a I addressed the comment in https://github.com/JMSLab/xtevent/issues/170#issuecomment-2035206257.

• After the changes, I ran the test file and found a conflict when estimating static models. The window parser was activated when the user asked for a static model. In https://github.com/JMSLab/xtevent/commit/4017f257e8d458f2e60f9b8be8ce681f63f71865 and https://github.com/JMSLab/xtevent/commit/80e8a07006ab8b9f1caeba9d8c7e47c79c3395ce I added some changes to overcome this conflict.

After this, I ran the test file and the file in the issue folder without inconvenience. I checked that the help file was rendered correctly as well.

[Constantino-Carreto-Romero]

Summary: In this issue, we implemented automatic window generation if the policy variable follows staggered adoption. Thread continues in https://github.com/JMSLab/xtevent/pull/178