Original: 5c929ea3f2d4e6ac28d18310d08474529eb2f71f

brought in through glue dependencies, this was done for the debian
packaging in a99092c2fff3c8e724f3e047a81d389fea34a0cb, this ports that
change to the SL packaging
Original: a20c2b70e0413f65b08d57f251057257a3bd9530

Original: 5a6c52aa86a5087b4b65e1f8752818afbb72e0c4

Original: 99baace9bf7fde4c8e85337bc1dae7984386b595

Original: 1162f19d6d600d737667e9c5e660c733933aae2a

Original: c384cf90107e212e637c75d0cf0c93a4f352934f

Original: b84aeb329dff762d1e070e8acb07c8cde579e87d

- type is being removed
Original: 31a906b4df995538a3fcd5801633ba3639ac17dd

Gstlal an PyCBC send us more data than we need. Trim the SNR time
series before passing to BAYESTAR in order to speed up time
marginalization.
Original: e39ed96eb8e21d63139c2182c78a768eb6137cad

This reverts commit 277dfbb4a53eada9605dbdff034ffa63d00b6bc1.
This change actually made it slightly slower.
Original: 57d3cd14a347531dd03a8d31787ce35067918819

Original: 277dfbb4a53eada9605dbdff034ffa63d00b6bc1

It cuts down on the integer-to-floating-point conversions.
Original: 2ab7db7b6d2b76ab22fc5b7099c3ca055622c56f

Original: f510674fda26cc97916c643ef31382bcb7a5e42a

Original: cb65a31f182e383ee4995370ce3645dcdcef0e84

Original: 0b441a187cc8ff463be158d338d93db50082846e

- Move it to its own compilation unit.
- Restore dynamic allocation init function.
- Precompute bounds checking.
- Precompute invalid value checking.
- Apply the same speedups to the 1D cubic interpolator.
- Expand unit tests.
Original: 12bbc4674e410f7cf2eeeb516bfe6eeafb1835cf

Original: 0d340a58fa964ca1ca2d5d4ccbab7732df623e2a

Original: 6c12124070c4c04efedacf23bf57e5cf4d989229

Original: e24fda05921694669c47394ed741465935e9d3bc

these are brought in through python-glue dependency
Original: a99092c2fff3c8e724f3e047a81d389fea34a0cb

Original: a717f1603512b297d1b8c499e870421c31591a03

Original: d945117824f482fd69d75da7e6a971b62fbb9d6d

Original: cbfc4d5bc082c73bc21cb27d4efd4eab5c6aa806

Original: b59e48fa6bc7c9c0d94feb848ab96246abd185fe

Original: 4c52c5a452c114016c8898d327aee48096bab26d

Original: bdefef2d4eefb1eb41e3e63c910364266325e1de

Original: 7bb4a2cfc1f97a2ac5564853b01e2d4b91d982e4

Original: 85a50c3bb5e2724668d3446ad2adf43756447fe5

Original: c440f89b75ad7b24b10b3a3a0d351d73561929fd

Original: 7edf99d912bba9f94b3ad8568a3bdbf5a846f3b4

Original: 89ef946c9da72ab36686c215a007e2b806db5e58

Original: 116fb529b26eaadb0f05f7680a9a42e958e8cc35

Original: 2ae5d47dc295bebcf75a85f9d3f93f935054721f

Original: b58e80212d9520568d44a96189ee3551d279a7d7

This speeds up BAYESTAR by about 50%.
Original: 61ef915ad390a9f8a51f7600561ce2fe714f9078

For large numbers of terms, logaddexp() performs about twice as many
transcendental function evaluations as is necessary.
This patch results in a speedup of about 25%.
Original: 41634d246522aefd772d5f6b19d13277476b9c07

Original: 43695a0b77d081dcb685d06e50d373bb418894ea

Original: 03127c12a26265143585fd0fc72261227f7720f9

Original: a41aabbf3570cb80a23c5bd277caf0bb8bc8900f

Original: 808c40b6c4fb2b5f9137d4ccf57f28d636b642fb