@vgalin review please

Thank you for taking the time to create this PR. Along with the (minor) reviews, I have a question myself:

What happens if a user is using an older version of Chrome(ium)? Does it break the headless mode because headless=new|old is used instead of just headless, or does it still work fine?

Regardless of the outcome, I see two possible courses of action:

• Do nothing: If someone is using an older Chrome version, they’ll need to use an older version of html2image as well.
• Allow the user to choose between headless=new, headless=old, and headless. Currently, the attribute you added sets headless to headless=new or headless=old based on its truthiness. Perhaps a third option could be introduced: if the attribute is None, then the value is simply headless without any suffix.

Do as you see fit, I don't have a preference at the moment. If I change my mind, I can always add the second option later.

So does this mean the browser_executable parameter is no longer supported? Is there an intended retrofit for those of us who were using that (to use Edge instead of Chrome)? Thanks!

@jrb-github Not exactly. This only concerns the CLI script of html2image, and browser_executable has been removed because it was wrongly used as a parameter of the screenshot method, which no longer supports this parameter.

If you're only using html2image through Python files/scripts and not through the CLI, nothing changes for you. If you're using the CLI, it will "just" become functional again.

@jrb-github Not exactly. This only concerns the CLI script of html2image, and browser_executable has been removed because it was wrongly used as a parameter of the screenshot method, which no longer supports this parameter.

If you're only using html2image through Python files/scripts and not through the CLI, nothing changes for you. If you're using the CLI, it will "just" become functional again.

Hm. Maybe I should branch this off into a separate issue, but I found this pull request because of a recent change in behavior where I was successfully using the browser_executable parameter in the Html2Image constructor (python code, not CLI), and when using the screenshot method on the resulting object, it no longer generates the correct image. Obviously I'm not using this code as the change hasn't been committed yet, but it seemed so closely-related in timeframe and behavior that I assumed it must be connected.

EDIT - I assume based on your comments that the behavior I'm seeing is unrelated. I submitted a new issue here.

@vgalin Thanks for the review!!! I believe the best way is to leave it backwards compatible anyway.
I made the changes as you suggested on 9e96083