exclude UTF-8 BOM character from aggregated js, css files

If you have, say, two css files,

h1 { font-family: sans-serif; }

and

body { background-color: green; }

and the second file is encoded as UTF-8 (with BOM), the character \0xFEFF is inserted before the letter b of body when css optimization is turned on, causing the optimized site to look different than the non-optimized site.

Perhaps this character could be intercepted and destroyed during the aggregation process.

See also the following which might be related:

http://drupal.org/node/661922
http://drupal.org/node/683712
http://drupal.org/node/361186

cheers,

Albert.