[Docs] Clarify "Large Projects" section

[gbrova]

This PR clarifies the documentation about large projects.

If my understanding is correct:

• Using slim_handler increases disk usage, not memory requirements.
• We are still limited to a maximum size of 512 MB, because AWS restricts the size of /tmp.

[coveralls]

Coverage remained the same at 72.806% when pulling 21a2dc0 on gbrova:gbrova/docs-large-projects into 4cf80b2 on Miserlou:master.

[coveralls]

Coverage remained the same at 72.806% when pulling 21a2dc0 on gbrova:gbrova/docs-large-projects into 4cf80b2 on Miserlou:master.

[cl1ent]

according to the docs, the limit for deployment package (zip) uploads from s3 is at 250mb (or more specific 262144000 bytes) and not 512mb. its even smaller for direct uploads.

see here and here

[gbrova]

Yes, that is true.

However, if my understanding is correct, the purpose of the slim_handler flag is to overcome that restriction.

Reading through #538, seems this is implemented by copying the zip to /tmp. So, we are still restricted on space, but the restriction is slightly more lax -- according to your first link and also the discussion in 510, the limit for /tmp seems to be 512MB.

It would be nice if this limit was documented somewhere, because I initially (incorrectly) assumed that I can deploy arbitrarily large functions using slim_handler.

Please let me know if I'm misunderstanding something.