…ible parsing (eg tolerant of arbitrary whitespace)

Commands report `ParseError`s if the syntax is invalid (eg too many or not enough args)
This required making `Resolver::add_code_source` public and adding `Context::resolver_mut` to track source code positions of commands

Added some helper functions for commands

…ands

Note that there is a bug where non-command inputs run `add_course_source` twice, which leads to incorrect line numbers

…s of it (much cleaner)

…and input

How: split `CommandParser` into a second struct, `SourcelessCommandParser`, that contains just the input, no `code_source_id`
`SourcelessCommandParser` has a fallible initializer that only returns `Some(Self)` if the line is indeed a command (ie starts with a command word)
Then we simply check if this initializer succeeded; if so then make a new `code_source_id` and construct the full `CommandParser`, then parse the command

…chable!`)

Made "list" report incorrect number of args before reporting invalid arg, in the event that both errors were present

Replaced `ensure_zero_args!` macro with function, now that this is possible (since we can use `format!` instead of `concat!)

…t them from being removed when removing trailing whitespace on save

… reading prelude

… `run_repl`

Therefore changed `parse_and_evalute` to return a struct of `control_flow, errored` to let session history know whether the command in question errored
Added tests to `session_history.rs`
Code seems overall not terrible now

…>` and renamed some items accordingly

…ands

Improved save logic: allowed for 0 arguments, in which case session history will be saved to history.nbt
Printed message on successful save

…cy with CLI

Updated web docs

…orphic but more self-descriptive)