make everything be repeated, the client will have a rough time but th…

[kevinkreiser]

…e files are smaller

this cuts the size quite substantially. if you remember from #49 you can see that the worst case tile had lots of large parts. with this change we could probably up the total number of segments per subtile. note that this will not decrease memory demands client side, it really only impacts the serialized pbf (it compresses better this way) once you deserialize it client side it will still be just as fat as before.

@drewda @dnesbitt61 @louh

here's the new sizes:

foo:bar$ ls -hal *.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 13:54 415.nex.0.gz
-rw-r--r-- 1 kkreiser kkreiser 7.3M Aug  8 14:04 415.nex.10.gz
-rw-r--r-- 1 kkreiser kkreiser 9.3M Aug  8 13:55 415.nex.1.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 13:56 415.nex.2.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 13:57 415.nex.3.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 13:58 415.nex.4.gz
-rw-r--r-- 1 kkreiser kkreiser 9.1M Aug  8 13:59 415.nex.5.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 14:00 415.nex.6.gz
-rw-r--r-- 1 kkreiser kkreiser 9.1M Aug  8 14:01 415.nex.7.gz
-rw-r--r-- 1 kkreiser kkreiser 9.1M Aug  8 14:03 415.nex.8.gz
-rw-r--r-- 1 kkreiser kkreiser 9.2M Aug  8 14:04 415.nex.9.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:54 415.spd.0.gz
-rw-r--r-- 1 kkreiser kkreiser 3.7M Aug  8 14:04 415.spd.10.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:55 415.spd.1.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:56 415.spd.2.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:57 415.spd.3.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:58 415.spd.4.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 13:59 415.spd.5.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 14:00 415.spd.6.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 14:01 415.spd.7.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 14:02 415.spd.8.gz
-rw-r--r-- 1 kkreiser kkreiser 4.7M Aug  8 14:03 415.spd.9.gz

[dgearhart]

:shipit:

[kdiluca]

:+1: