The SAL should not need to curate stack IDs

There are several possible solutions.
- Linker tricks: each stack should place a symbol in a specific linker section. The index of the symbol is the stack ID
- String match: each stack defines an identifier string. The stack ID is the value of the pointer to the identifier string. Fallback is string comparison.
- yotta tricks: Similar to linker tricks, but instead yotta assigns a unique number to each module at build time. That number is the stack ID.

As user/partner-supplied stacks increase in number (e.g. due to external network modules) this will become more of a problem. One of the above solutions would be more effective.

No files or tests are named; begin by tracing how the SAL currently curates stack IDs and how user- or partner-supplied stacks are registered. Compare the linker-section, identifier-string, and yotta-assigned-number approaches, then define completion as removing manual stack-ID curation for externally supplied stacks.

Written by the indexing model from the issue text.