Open the README with the problem, not the description

[vahid-ahmadi]

The README opened with:

microdf provides MicroDataFrame and MicroSeries classes that extend pandas functionality with integrated weighting support, essential for accurate survey data analysis.

That tells a reader what the package contains, not why they would want it. The JOSS review checklist asks directly whether the software has a clear statement of need, and this is the first thing a reviewer reads.

Replaces the ## Overview paragraph with a ## Why this exists section stating the two problems the package actually solves, which are the same two the paper leads on:

1. The estimators are not the obvious ones. A weighted median is not the median of weighted values; weighted variance requires choosing between frequency and precision weights; a top-1% share requires deciding what happens to a record straddling the cutoff. microdf makes each choice once, documents it and tests it.
2. Weights have to survive the pipeline. Through merges, filters, grouping and reindexing, before any estimator runs. When they do not, nothing raises — the pipeline completes and returns a plausible wrong number.

Nothing else in the README changes: the features list, installation and usage sections are untouched.

[juaristi22]

Checked the two convention claims against the source. var uses sum(w) - ddof as the denominator and matches np.var on the replicated sample with integer weights. quantile is the inverse CDF, the smallest value where the cumulative weight share reaches q, which is what the docstring cites survey::svyquantile for. Both hold numerically.

The README is also the PyPI long description, so this shows up there at the next release. The fragment is .changed, so a patch bump. Approving.