What Zarx said (but as I was trying to recall, I thought MP4Box would give you the proper framerates automatically even if you just specify 23.976 for instance).
As for the resizing, yes and no. Depends how anal you want to be, and also your opinion/assumption on how the source was encoded.
See the SARs you specified are correct, they both produce a 16:9 frame shape on playback, be it originally 704x480 or 720x480. However what you need to bear in mind is that the majority of encodes that are 720 wide, have up to 16 pixels of dead space, or rather we can simply say it has an active area of ~704 pixels (samples is actully more accurate, but seeing as it's digitized).
You see where 873 comes from is Target active area size/Actual active area size*Overall frame size=Size that the actual active area should be resized to.
Ok, well that's not the clearest example ever, but lets fill in the values
853.33/704*720=872.73
When the 720 with borders is resized to 872.73, the active image without borders is now 853.33 wide.
Here is a little example.
First Image: 720px (704 active + 16px borders) resized to 873px
Second Image: 704px (all active) resized to 853px
Third Image: 720px (704 active +16px borders) resized to 853px.
Now you will notice that the first and second images are similar, give or take a line or two (due to rounding). Although the first image is stretched more, it's active area of 704px actually becomes in the region of 853px, which is the same as cropping the borders and resizing the full 704px to 853px.
Resizing the full 720px > 853px is a different story. Now the active area is only approximately. The active area would become 834px.
Most people will crop to 704 and set a 40:33 SAR (to get 853). Encoding it as 720 with borders and setting the appropriate SAR to get 873 essentially gives you the same results as just cropping to 704 and resizing to 853 (actually it's not as good because you are wasting bits on coding the borders).
Sometimes the borders are there to assist the correct AR (because MPEG-2 has a very limited set of aspect ratios, so adding borders or letterboxes help get the correct AR even with a limited set), such as excessive borders. This is where you have to use your own judgement and any information about the source.