Merge README into README.md and the manual

[fingolfin]

The old README file, written by the original authors, sat next to
README.md with no clear division of labour: parts of it duplicated the
manual, parts predated it, and parts had gone stale. This merges it away, so
that README.md is the only README.

Merge README into README.md and the manual

Its description of how the library is organised — the layers, their file
layout, the storage formats, and the identification algorithms — becomes a
new manual chapter, The Organisation of the Small Groups Library. The
[1]…[17] reference list turned into proper citations; most bibliography
keys already existed, and I added four (Hölder 1893 and 1895, Bagnera 1898,
James–Newman–O'Brien 1990). GrpConst and ANUPQ are now named as packages
rather than cited.

Two things were repaired along the way. The "Files and Directories" section
still described the old small directory of the GAP distribution, with
readsml.g and small.gi at top level; it now matches the actual package
layout. And doc/overview.xml pointed readers at "an extensive README
file … in the small directory of the GAP distribution", which now points
at the new chapter instead.

The introduction moved into README.md, with the list of available orders
refreshed to match the manual's rather than the 2008 one. Dropped entirely:
the authors' postal addresses, the "Available functions" list (the manual
documents these, and the copy in README was missing everything added since
2008), and the request to report bugs by e-mail — README.md already points
at the issue tracker.

Mention SglPPow and SOTGrps

A "Related packages" section in both the manual and README.md mentions
SglPPow (p^7 for p > 11, and 3^8; it extends this library, so its groups
arrive through SmallGroup and friends) and SOTGrps (orders factorising
into at most four primes, and p^4·q; its own functions, and its own
numbering, which differs from ours where the ranges overlap).

That covers the documentation half of #22 only — whether to also emit info
messages is still open there.

Don't duplicate the list of layers

The overview carried an abridged copy of the layer list. Now that the new
chapter says which orders each layer holds, the overview just links there.

Polish

The order bounds are spelled out as "at most 1000" rather than "<= 1000", as
the overview list used to do. And the paragraph claiming the data occupies
about 30 MB and the identification about 47 MB is gone: those figures
predate the compression in 1.5.1, and nobody counts megabytes for this
nowadays. It took the link to http://www.icm.tu-bs.de/ag_algebra/software/small/
with it — that page is still live, but its six "selected publications" are
all cited a couple of sentences further down anyway.

Resolves #35

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.84%. Comparing base (a172878) to head (b6f73ce).

Additional details and impacted files

@@           Coverage Diff           @@
##           master      #70   +/-   ##
=======================================
  Coverage   99.84%   99.84%           
=======================================
  Files         493      493           
  Lines      327377   327377           
=======================================
  Hits       326861   326861           
  Misses        516      516           

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.