Correct RECO output timestamps and complex metadata

[headmeister]

Summary

• correct RECO_abs_time to a Unix-format integer across all ParaVision generations
• distinguish RECO reconstruction-start timestamps from PV6+ VISU pvtime_t fields
• clarify scaled-intermediate minima/maxima
• document that complex imaginary frames reuse the real block's scaling and extrema metadata

Verification

Compared against PV5.1, PV6, PV7, and every available PV360 1.0–3.7 parameter manual. Documentation-only change; no manuals, inventory files, or unrelated ignore changes are included.

[gdevenyi]

Review against the reference material in resources/.

Verified: RECO_time is char[24] in PV5.1 and pvtime_t in PV6.0.1 (acq_extern.h); RECO_abs_time is int then pvtime_t; public files agree; the complex-image reuse sentence is verbatim in D13/D02/PV360; the RECO output arrays are not doubled (Zenodo 4522220 expno 8 pdata/2: ( 1 ) with two frames). Correct.

Contradicted by the public data: "stored PV6.0.1, PV7.0.0, PV360 3.6 and PV360 3.7 reconstructions all pair an ISO-form RECO_time with a three-field RECO_abs_time". The derived DTI pdata/2 reconstructions in the public PV360_StdData (github.com/cecilyen) and in Bruker's 360.3.5–3.7 standard datasets have a struct RECO_abs_time and no RECO_time; Zenodo 4522220 expno 43 pdata/1 has neither.

Notes: D02 and the PV360 manuals still print "char array" and "int … Unix format" for these two — worth one sentence that the headers and files win. The PR body says the opposite of the diff ("correct RECO_abs_time to a Unix-format integer"); only the diff matters.

Question: reword to "reconstructions that write RECO_time pair it with a three-field RECO_abs_time; derived PV360 DTI processings omit RECO_time"? Then I merge.

[headmeister]

Rebased onto current main and made the timestamp fields explicitly optional. The text now says that reconstructions writing RECO_time pair it with RECO_abs_time, while derived PV360 DTI processings may retain only RECO_abs_time and some processings omit both. It also records that the PV6.0.1 headers and files supersede the manuals older char/int descriptions. Across 165 mounted reco files: 135 contain both fields, 13 only RECO_abs_time, and 17 neither.

[gdevenyi]

Revision checked. The pairing statement is now conditional ("reconstructions that write RECO_time pair it with RECO_abs_time"), the PV360 derived-DTI omission and the both-absent case are stated, and the manuals' stale "char array"/"int" descriptions are noted against the PV6.0.1 headers. Matches the public data. Merging.