Describe the bug
Autocrafting pipe network overproduces subproducts. Or, rather, it doesn't count overproduction of subproducts, when >1 final result is needed.

To Reproduce
Steps to reproduce the behavior:

1. Build the following autocraft network:
   1. gold ore -> gold ingot
   2. autotable (ingot -> nugget, reeds -> paper, stuff -> null chip), with broadcast
2. put alot of ore, redstone and reeds into linked chest
3. ask system for 64 null chips
4. observe that 477 gold ingots were produced as well
5. subproducts are in autotable, so unless broadcasted the system doesn't know they are even there

Expected behavior
I expected the amount of subproduct being below their batch-of-production numbers (< 9 nuggets, <3 papers, etc).

My thoughts
Probably, using autotable for > 1 recipe is the cause of the problem.

Crash Log
None.

Versions
Include versions of the dependencies. And since you're looking at it, make sure ProjectRed is up to date. Don't submit bugs for old versions, as they could have been fixed.

• CCL Version: CodeChickenLib-1.12.2-3.2.3.358-universal
• FMP Version: ForgeMultipart-1.12.2-2.6.2.83-universal
• MrTJPCore Version: MrTJPCore-1.12.2-2.1.4.43-universal
• ProjectRed Version: ProjectRed-1.12.2-4.9.4.120-*

Screenshots/Videos
If applicable, add screenshots or video links to help explain your problem.

After you request 64 null chips next time it will use these overproduced dependencies, though. However, producing that much nuggets can overflow the storage and in case you request some "solar panel XIII" leave you with storage filled with 2^12 of some "solar mirrors" needed for "solar panel I", since I assume the same sequencing is used for constituents of recipe.

This can explain 477 nuggets: there are 2 in null chip, so for each one an ingot is requested.

Probably, when batch-producing the algorithm does not account for subproducts left from previous item.