Document the recommended way(s) to run standalone bots

[noemartinez]

The only way of running bots currently captured in the docs is to clone python-zulip-api, put the bot in zulip_bots/zulip_bots/bots and run it by name using zulip-run-bot.

For quick bot prototyping, and for writing bots that are meant to always remain standalone (not meant to be contributed back to or deployed as part of python-zulip-api), it's more practical to run the bot using the existing zulip-run-bot support for passing an explicit path to the bot's handler .py file. And soon, there might be a third supported way to run standalone bots (see #357).

Could we add that/those zulip-run-bot workflow(s) to the documentation to highlight a way to run bots which doesn't require to clone python-zulip-api or store the bot code in zulip_bots/zulip_bots/bots?

I'd personally argue this should even be the main, simplest, documented way, leaving something like "Contribute your bot to Zulip" (i.e. python-zulip-api) as a separate, secondary section highlighting the zulip_bots/zulip_bots/bots workflow as an opt-in thing a developer can choose to do. But that's just my two cents. As long as it's documented, that's the main request here :).

See this conversation for more context on this.

[timabbott]

See also #356 as something we might want to document in the meantime.