I added the lf-no-validate class to every hidden field.  

Many of these are hidden via a class of hiddenField added to the fields in the Layout screen.  The CSS code for that is:

.hiddenField {display: none!important;}

I added the lf-no-validate class to all of those fields with this Javascript code: 

//Set the hiddenField items to be excluded from field validation
$('.hiddenField input').each(function() { 
  $(this).addClass('lf-no-validate');
});
$('.hiddenField textarea').each(function() { 
  $(this).addClass('lf-no-validate');
});
$('.hiddenField select').each(function() { 
  $(this).addClass('lf-no-validate');
});

That adds the class to every hidden field that started with the hiddenField class - input catches text fields, dates, checkboxes, etc. - textarea catches multi-line text boxes - select catches dropdown fields.

I also have fields that are hidden/shown in my Javascript.  These are hidden/show by adding or removing the hiddenField class throughout my Javascript.  I just made sure that every time I hide a field, I add the lf-no-validate class to its input/select/textarea element as well.  And every time I show a field, I remove the lf-no-validate class from its input/select/textarea element as well.

Now when I click submit, it takes two seconds instead of a minute or more.  Yay!

I have not yet experimented with doing an alternate validation.  My form uses pagination, with tabs, and all fields that need validation are contained in collections on each tab/page - so my idea would be to disable all the tabs initially, and then on each page, complete validation of that page's collection when they click the Next button, if the validation doesn't pass it stays on that page, if it does pass, that page's tab will be enabled and the user will go to the next page.  By disabling all the tabs and enabling them one at a time, the user would be able to go back to tabs they had been at before, but not skip past any tabs that had not been through the validation process.  I may still do this, but for now, my form is running efficiently enough with the other changes, that it isn't necessary.

Thanks again @████████ for your help!