fix: rely on JuliaSyntax's tokenize instead of regex-based splitting of strings during conversion to expr

[sstroemer]

Fixes #35

This attempts to (sustainably) fix the error that occurred due to the regex-based splitting of strings during "to expression" conversion. It now properly tokenizes strings. However, that also revealed that column names like 08_pv@data are ambigous, since that would be valid Julia syntax for 8 * _pv@data. If this is detected the model building is aborted (since we don't know what a user actually means).

The MWE from #35 is now included as test case, and a base set of reasonable unit tests has been added.

---

This pull request includes updates to various example configuration files to ensure compatibility with version 2.0.3 of the core library. Additionally, it updates the references to data columns in several examples to reflect new naming conventions.

Data column reference updates:

• Updated availability_factor and value references in src/assets/examples/07_csv_filestorage.iesopt.yaml to use the new naming convention. [1] [2]
• Updated availability_factor and value references in src/assets/examples/08_basic_investment.iesopt.yaml to use the new naming convention. [1] [2]
• Updated availability_factor and value references in src/assets/examples/10_basic_load_shedding.iesopt.yaml to use the new naming convention. [1] [2]
• Updated conversion and value references in src/assets/examples/15_varying_efficiency.iesopt.yaml to use the new naming convention. [1] [2]
• Updated capacity reference in src/assets/examples/17_varying_connection_capacity.iesopt.yaml to use the new naming convention.

[sstroemer]

CI runs are manually started here: https://github.com/sstroemer/IESopt.jl/actions/runs/12058593193

[sstroemer]

80f7605 fixes a spelling mistake (not checked again in CI; not sure why my pre-commit did not catch that locally)

[sstroemer]

This already includes a full bump to v2.0.3, which we then need to replicate over at iesopt.