Docs for how to start a sandbox node and run e2e test on it

[ailisp]

Add a doc on how to start a sandbox node and how to run e2e test on it. It's going to supersede end-to-end test sections once our tooling is fully ready that support run test on both sandbox and testnet with one cli command.

Update:
added a subsection of how to use sandbox-only feature (right now there's only one feature: patch_state)

[render]

Your Render PR Server URL is https://docs-pr-706.onrender.com.

Follow its progress at https://dashboard.render.com/static/srv-c2jv9n32v8panf9vsk2g.

[matklad]

cc @DiscRiskandBisque (am I correct that you do want to be cc'ed for such design/UX aspects of this work?)

[DiscRisknBisque]

cc @DiscRiskandBisque (am I correct that you do want to be cc'ed for such design/UX aspects of this work?)

Yes and thank you! Apologies for the delay, I need to set up better GitHub notifications.

[thisisjoshford]

Left some grammar edits, suggestions, and comments :)

[matklad]

Finally got to reviewing this, sorry that it took me so long.

This looks good to me overall.

However, I am not sure what are concrete next steps here. The current state is very a proof-of-concept. It works, but the UX is pretty horrible in itself, and we don't give any explicit guidance for contract authors about the interactions between this new style of testing, and existing simulatior/unit-test.

As such, I don't think we should be publishing this as a user-visible documentation, as it seems that that'll just create confusion over currently supported testing approaches.

It seems like this is the point where the work should be taken over by someone from the devx team, but I don't know who that would be, cc @mikedotexe.

[ailisp]

@matklad 's point makes sense too. My opinion is present both manual, low level way and a high level, automatic way to the user, with advanced user might be interested to learn when high level way doesn't work, e.g. when the automatic way has a bug or a special case that automatic way doesn't cover. Right now we don't have truffle equivalent so this doc only has the low level way. Once we have, then the high level way should be add to the first part as of this document so user can see the short way first.

The other PR that documents repro testnet transaction with sandbox have applied this approach: https://github.com/near/docs/pull/724/files. First sections give the instructions on how to use a automatic script to fetch state for reproduce. The later section describes which parts of blockchain information will affect the transaction result and may help curious user to understand the low level and troubleshoot if it there's still difference in transaction repro.

[ailisp]

Seems latest master d894e82 broke CI / test-links, let me see if i can make a fix to master.
Edit: tried to debug it a little big, however the build log on github actions is not informative, and also i'm unable to reproduce the failure locally (i'm in china behind a vpn but this test-link.sh doesn't seem to pickup my system proxy settings so failed at very first links of https://docs.google.com, unable to see what's actually failed), Maybe @mikedotexe can take a look?

[ailisp]

This one is now blocked by near/near-sdk-rs#433

[ailisp]

@thisisjoshford would need your or Mike's approval on deprecated-near/rust-status-message#64, thanks