Is it just because it's not semantically correct? Neither is a grid system.
Is it better for responsive designs?
<table width="100%">
<tr>
<td width="50%">Column 1</td>
<td width="50%">Column 2</td>
</tr>
</table>
Compared to using a grid system (12 columns for the sake of example):
<div class="container">
<div class="span6">Column 1</div>
<div class="span6">Column 2</div>
</div>
With the table, well, you're out of luck. The table markup determines layout, and you can't reflow table columns to make them table rows using CSS. I suppose you could use some JavaScript to rewrite the table on the fly, but just using a grid in the first place seems a lot easier.
Another benefit of a grid is that if you do care about semantics, you don't have to use meaningless divs: you can use any block-level element. This would work just as well as the grid example above:
<body class="container">
<article class="span6">Column 1</article>
<aside class="span6">Column 2</aside>
</body>
A third example: let's say you have the two column layout, but you want to dynamically add or remove the second column. If you're using a grid, no problem: the second column can be hidden or shown without affecting the first column's width.
If you're using a table, the first td will reflow because table columns will take up the entire width of the table. To get the same effect, you'd either have to adjust the width of the table or remove the contents of the second column's td instead.
The above also applies if you just want to change the width of the second column: if, say, you wanted to make the second column 25% width and leave the first at 50%, it's no problem with a grid: just change the second column's class to "span3". With a table, if you change the second column's width to 25%, the first column will expand to take the rest of the table's width. You'd either have to add a third td as a placeholder or resize the table itself.
There are a lot of use cases like these where having a grid winds up being a lot simpler than using a table. The benefit of a table-based layout was that creating a cross-platform grid was hard, and there wasn't a whole lot of prior art, so a lot of times you might have to start from scratch. But if you already have a grid-based system available, there's not much point to forgo that and use a table.
Tables are especially helpful for creating a horizontal list of elements that are evenly spaced, but of uneven/unknown width. You can't do that with a grid. Since grids are floated, it's easy to adjust the number of elements in a row (by either changing the width or the row or the width of the items). You can't do that with a table, since each row has to be delineated in the markup.
I don't use table elements for layout of course, I use display:table. Using a table element for non-tabular data would be unsemantic, unlike using a div which carries no semantics.
But hey, "they're' not tables - tables are evil!", right?
RANT: I worked with some people ... sheesh - almost a decade ago that proclaimed "tables are depreciated" (I'm pretty sure they meant 'deprecated') and spent days having a team of people convert tablular data from Excel spreadsheets in to bloated DIV-hell for IE5 and NS4 browsers. That's been my worst experience with table-haters, but it stuck with me.
(Look at HN's code source)
Also it's worth noting that most codes using Foundation or Bootstrap are considered bad code for the reason I just told you : You're not supposed to have elements of design in your html page.
The correct way of using Foundation/Bootstrap would be to use mixins directly in the .scss files of your site (if you're using foundation, bootstrap should use LESS files and I have no idea how it works).
For what it's worth, I tried searching for "semantic grid system" and found http://semantic.gs. I don't really know if it's any good but I like the idea.
And since it isn't a table of data, the grid is more semantically correct. Not because the grid has any semantics itself, but because you're not misusing other semantics.