Hash every file's head, tail and content in the hash phase (closes #132)

Every file the walk records now gets three hashes in the hash phase:
head and tail, of its first and last 64 KiB, and content. The content
hash covers the whole file under 100 MiB, and 1 MiB at each gigabyte
from 100 MiB up, so a large file is never read in full. One 100 MiB
constant replaces the 10 MiB and 50 MiB thresholds. The size census,
the records stored without hashes and the content phase are gone, so a
scan has four phases and a rescan reads only new and changed files.
report and trees no longer special-case a missing content hash. README
"Duplicate detection" and "Rationale" are rewritten.

Judgement call: info drops its two hash-count lines, which would now
always equal the record count.

Model: opus-5-5

README.md

@@ -272,7 +272,7 @@ Four subcommands, all implemented:
     from the database records, compute a Merkle-style digest per directory, and
     report maximal groups of identical trees.
 4.  `info` — the database's details: its path, the sizes of its files, its
-    schema version, the number of its records and their total size.
+    schema version and the counts of its records.
 
 ```
 sfdupes scan [--workers N] [-x] [--no-default-excludes] PATH...
@@ -427,8 +427,8 @@ The content hash depends on the file's size:
   of identical content, barring a SHA-256 collision.
 - At 100 MiB and above, it is the SHA-256 of samples, so the file is never read
   in full. A sample is the 1 MiB at each gigabyte-aligned offset: 0, 1 GiB, 2
-  GiB and so on. The samples are fed into one hash in order. A sample that would
+  GiB and so on. The samples are fed into one hash in order, and the last one
-  run past the end of the file stops there.
+  stops at the end of the file.
 
 The sampled hash is **deliberately probabilistic**. Bytes outside the samples
 and outside the first and last 64 KiB are never read. Two large files that
@@ -912,15 +912,15 @@ Additional requirements:
   terminal is. A line too wide for the terminal first loses its bar, which gives
   way to the spinner, then its rate, the words of a stalled phase, or what the
   phase is waiting for. The phase and its place, the progress, the elapsed time
-  and the ETA are kept whole. On an 80-column terminal, at counts like those in
+  and the ETA are kept whole. On an 80-column terminal, a stalled line keeps its
-  the example above, a stalled line keeps its words whenever the elapsed time
+  words whenever the elapsed time and the ETA are both in minutes. It holds at
-  and the ETA are both in minutes. Where the words and the ETA do not both fit,
+  counts like the example above. Where the words and the ETA do not both fit, as
-  as at millions of files and times in hours, the words give way. Only when
+  at millions of files and times in hours, the words give way. Only when those
-  those four alone are too wide, as on a terminal narrower than about 70 columns
+  four alone are too wide, as on a terminal narrower than about 70 columns hours
-  hours into a scan of millions of files, is the line cut short at the
+  into a scan of millions of files, is the line cut short at the terminal's
-  terminal's edge, so the end of the ETA is lost first. A terminal made narrower
+  edge, so the end of the ETA is lost first. A terminal made narrower while a
-  while a phase runs gets the same: the next draw erases the line within the new
+  phase runs gets the same: the next draw erases the line within the new width
-  width and fits the line to it, and the display goes on redrawing in place.
+  and fits the line to it, and the display goes on redrawing in place.
 - When stderr is not a terminal (a pipe, a file, `/dev/null`), do not emit ANSI
   redraws: print the line the moment each phase starts, every 5 seconds while it
   runs (whether or not an item has completed, and never more often), and once