Waste of what?
> Because if not every font implements it, what is the usefulness of it?
Most fonts only implements a fraction of unicode, is unicode a waste? Is cyrillic in unicode a waste? Is Cherokee? Braille symbols? Buginese and Meroitic cursive?
> Did you ever visit websites that use glyphs your font doesn't know?
Yes? The software will fall back on a font which does provide the glyphs, or display a block with the codepoint if it doesn't have any.
> How would you use those pictures on a website, when you can't even guarantee the style of the picture, because it may be implemented differently or not at all.
How would you use letters on a website, when you can't even guarantee the style of the letters, because it may be implemented differently or not at all?
Not only that, but unicode is not specific to websites for fuck's sake, not only will somebody find a use for them on websites even if you can't, somebody will definitely find a use for them outside of websites. Most (or all?) of the new emoji are inherited from windings[0] and webdings[1], they don't come out of nowhere.
> It just looks like a big waste for no good reason.
Again, a big waste of what?
No more so than right now.
> selecting a reasonable subset of bazyllion random, singular pictures to draw and test is a major headache.
Right, because there[0] were[1] no[2] "bazyllion[3] random[4], singular[5] pictures[6] to[7] draw[8] and[9] test[10]" dating back to Unicode 1.0[11] before and Unicode 7.0 changes everything.
[0] http://en.wikipedia.org/wiki/Mathematical_alphanumeric_symbo...
[1] http://en.wikipedia.org/wiki/Mahjong#Unicode
[2] http://en.wikipedia.org/wiki/Dominoes#Dominoes_in_Unicode
[3] http://en.wikipedia.org/wiki/Playing_Cards_(Unicode_block)
[4] http://en.wikipedia.org/wiki/Musical_Symbols_(Unicode_block)
[5] http://en.wikipedia.org/wiki/Arrow_(symbol)#Arrows_in_Unicod...
[6] http://en.wikipedia.org/wiki/Optical_Character_Recognition_(...
[7] http://en.wikipedia.org/wiki/Geometric_Shapes
[8] http://en.wikipedia.org/wiki/Unicode_Dingbats
[9] http://en.wikipedia.org/wiki/Phonetic_Extensions
[10] http://en.wikipedia.org/wiki/Miscellaneous_Technical_(Unicod...
[11] http://en.wikipedia.org/wiki/Box_Drawing_(Unicode_block)
Other comments have touched on this already, but I want to substantiate with additional information: In practice, fonts that cover many different writing systems at once are relatively rare. Your typical Western font will focus on Latin with varying coverage of descendants (like the Portuguese and Vietnamese alphabets) and perhaps expand to cover Cyrillic, but not more. Conversely, it's rare for a font by a Korean foundry to have equivalent Latin/Cyrillic coverage next to its native Hangul alphabet.
There's plenty of documents that mix these writing systems, though. I do in fact visit websites that aren't covered by any one font in my browser's default settings or any one family named in the site's markup/CSS all the time, and you probably do too.
Hence the font stacks on all relevant platforms perform glyph substitution: If font X doesn't contain a glyph, it'll go looking for a font Y that does and drop that glyph in. Many platforms allow you to manipulate substitution preference e.g. by defining aliases for font families.
It just looks like a big waste for no good reason.