Add more ERP measures

[BelizSertcan]

This PR completes @withmywoessner 's work in #12434.

Reference issue (if any)

Closes #7848
Closes #12434

What does this implement/fix?

Implements the following ERP measures:

• Mean amplitude
• Area amplitude (all positive, numeric integration)
• Fractional area latency
• Fractional peak latency

Changes relative to #12434:

1. Following the discussion in the previous PR, average=True now returns a
   single-row DataFrame, with the channel index shown as "average", so the
   column layout stays consistent with average=False.
2. Renamed channels to channel for consistency with the rest of MNE.
3. Replaced the original tests, which only asserted DataFrame.shape. The new
   tests check computed values against closed-form results.
4. Each function calls _check_pandas_installed(strict=True) rather than
   importing pandas at module level, so pandas stays optional for compute_sme.

Bugs fixed from #12434:

1. Unit mismatch in compute_frac_area_latency.
2. compute_frac_peak_latency returned the peak's own latency for "offset"
   rather than the true post-peak crossing.
3. All four functions crashed with any picks other than "all" — the channel
   column used the full ch_names while the data was the picked subset. Now
   resolves picks via _picks_to_idx before building the labels.

Additional information

Question for reviewers:
Should compute_frac_area_latency support mode='intg'? For a biphasic
signal the lobes can cancel, so the cumulative integral isn't monotonic and
"the latency at which X% of the area is reached" may have multiple solutions
or none.

Claude Opus 5 helped identifying the bugs and writing the tests, and drafted this PR description.

[larsoner]

You'll need similar changes to other functions. Anything following the : in param names needs to resolve to a intersphinx-findable cross-ref entry, or have an ignore in doc/conf.py:numpydoc_xref_ignore. You will see CircleCI complain about these formatting and reference problems...

[welcome]

🎉 Congrats on merging your first pull request! 🥳 Looking forward to seeing more from you in the future! 💪

[larsoner]

Looks great, thanks @BelizSertcan !