Add Bruker TopSpin NMR reader

What this PR does

Adds a reader for Bruker TopSpin NMR experiment folders so that solution-NMR data can be loaded into ixdat.

from ixdat import Spectrum

spec = Spectrum.read("path/to/topspin_experiment_folder", reader="bruker")

spec.plot()                    # intensity vs. chemical shift (ppm)
print(spec.metadata["NS"])        # number of scans
print(spec.metadata["SOLVENT"])   # e.g. "Urine"

spec is an NMRSpectrum (a new technique class added in this PR) and behaves like any other ixdat Spectrum.

The "TopSpin experiment folder":

The folder produced by Bruker's TopSpin NMR software. Typically looks like this:

my_experiment/
├── acqus            # acquisition parameters (text, JCAMP format)
├── fid              # raw time-domain data (binary)
├── pulseprogram
└── pdata/
    └── 1/
        ├── procs    # processing parameters (text)
        ├── 1r       # processed real spectrum (binary)
        └── 1i       # processed imaginary spectrum (binary)

You point the reader at the top folder (my_experiment/) and it figures out the rest.

How it works

The reader uses [nmrglue](https://nmrglue.readthedocs.io/) (Python package for NMR file I/O) to do the actual binary parsing. ixdat then wraps the result into its standard Spectrum data model:

• Processed data: reads pdata/<procno>/, returns the real spectrum with a proper chemical-shift axis in ppm.
• FID fallback: if no processed data is available (or you pass prefer_processed=False), the reader returns the magnitude of the raw FID against an index axis. Phasing/Fourier transform/referencing is left to the user.

nmrglue is not added to ixdat's required dependencies, it's only imported lazily inside BrukerNMRReader.read(). If a user tries to read a Bruker folder without it installed, they get a clear ReadError explaining how to install it (pip install nmrglue).

Metadata

All the key acquisition parameters from acqus are moved onto spec.metadata with their original Bruker names (PULPROG, SOLVENT, NS, DS, TE, BF1, SFO1, O1, SW, TD, NUC1, DATE, ...), and processing parameters from procs are exposed under proc_* prefixes (proc_SI, proc_SF, proc_OFFSET, proc_LB, proc_PHC0, ...). The full acqus and procs dictionaries are also preserved under spec.metadata["acqus"] / spec.metadata["procs"] so nothing is lost. Everything in metadata is JSON-serialisable.

The acquisition DATE is parsed into ixdat's standard unix timestamp (spec.tstamp) so NMR data can be co-plotted on a shared time axis with other techniques.

New technique class

Adds NMRSpectrum and NMRSpectrumSeries in ixdat.techniques.nmr, following the same pattern as the existing FTIRSpectrum / FTIRSpectrumSeries. Registered as "NMR" and "NMR_spectra" in TECHNIQUE_CLASSES.

Tests

A new tests/functional/test_bruker_reader.py is added.

The tests run against a real TopSpin 1D ¹H NMR dataset vendored under test_data/bruker/MTBLS1_ADG19007u_162_10/. It comes from MetaboLights study MTBLS1 (a publicly availiable human-urine 1H NMR dataset, 700 MHz, noesypr1d, 128 scans), MIT-licensed reference repository. I included full attribution in test_data/bruker/MTBLS1_ADG19007u_162_10/CREDITS.txt. Only the files strictly needed by the tests are vendored (~770 KB total).

Demo script

development_scripts/reader_demonstrators/demo_bruker_reader.py :

1. Reads the test-data via Spectrum.read(..., reader="bruker").
2. Prints the basics (class, technique, name, tstamp, ppm range, point count, intensity range).
3. Prints the lifted acquisition parameters (PULPROG, SOLVENT, NUC1, BF1, NS, TE, ...) and processing parameters (SI, SF, OFFSET, LB, PHC0, ...).
4. Plots intensity vs. chemical shift with the conventional inverted ppm axis, titled with the nucleus / pulse program / field / number-of-scans / solvent.

Files added / changed

New:

• src/ixdat/readers/bruker.py: BrukerNMRReader implementation.
• src/ixdat/techniques/nmr.py: NMRSpectrum / NMRSpectrumSeries.
• tests/functional/test_bruker_reader.py
• development_scripts/reader_demonstrators/demo_bruker_reader.py: demo
• test_data/bruker/MTBLS1_ADG19007u_162_10/: test dataset

Modified:

• src/ixdat/readers/__init__.py: registers "bruker" in SPECTRUM_READER_CLASSES.
• src/ixdat/techniques/__init__.py: registers "NMR" and "NMR_spectra" in TECHNIQUE_CLASSES.
• README.rst: adds an "NMR" row to the techniques/readers table. (EDIT: Accidentally removed the nordic-reader line and then added it back by mistake on main-branch; so we have patched it back here from main to not have merge-issues)
• NEXT_CHANGES.rst: entries describing the new reader and the new technique classes.

Things out of scope for now:

• 2D and higher-dimensional NMR experiments. This reader is scoped to 1D experiments for the time being. Support for 2D could be a future addition.
• FT/phasing/referencing of raw FIDs. The FID fallback intentionally doesn't try to do this; users who need that should preprocess the data in TopSpin (or with nmrglue) and let ixdat read the processed spectrum. We could eventually incorporate this as something like a new Calculator